001package org.jsoup.helper;
002
003import org.jsoup.Connection;
004import org.jsoup.HttpStatusException;
005import org.jsoup.Progress;
006import org.jsoup.UnsupportedMimeTypeException;
007import org.jsoup.internal.ControllableInputStream;
008import org.jsoup.internal.StringUtil;
009import org.jsoup.nodes.Document;
010import org.jsoup.parser.Parser;
011import org.jsoup.parser.StreamParser;
012import org.jspecify.annotations.Nullable;
013
014import javax.net.ssl.SSLContext;
015import javax.net.ssl.SSLSocketFactory;
016import java.io.BufferedInputStream;
017import java.io.BufferedReader;
018import java.io.BufferedWriter;
019import java.io.ByteArrayInputStream;
020import java.io.IOException;
021import java.io.InputStream;
022import java.io.InputStreamReader;
023import java.io.OutputStream;
024import java.io.OutputStreamWriter;
025import java.io.UncheckedIOException;
026import java.net.CookieManager;
027import java.net.CookieStore;
028import java.net.InetSocketAddress;
029import java.net.MalformedURLException;
030import java.net.Proxy;
031import java.net.URL;
032import java.net.URLEncoder;
033import java.nio.Buffer;
034import java.nio.ByteBuffer;
035import java.nio.charset.Charset;
036import java.nio.charset.IllegalCharsetNameException;
037import java.nio.charset.StandardCharsets;
038import java.util.ArrayList;
039import java.util.Collection;
040import java.util.Collections;
041import java.util.LinkedHashMap;
042import java.util.List;
043import java.util.Map;
044import java.util.concurrent.locks.ReentrantLock;
045import java.util.regex.Pattern;
046import java.util.zip.GZIPInputStream;
047import java.util.zip.Inflater;
048import java.util.zip.InflaterInputStream;
049
050import static org.jsoup.Connection.Method.HEAD;
051import static org.jsoup.helper.DataUtil.UTF_8;
052import static org.jsoup.internal.Normalizer.lowerCase;
053import static org.jsoup.internal.SharedConstants.DefaultBufferSize;
054
055/**
056 * Implementation of {@link Connection}.
057 * @see org.jsoup.Jsoup#connect(String)
058 */
059@SuppressWarnings("CharsetObjectCanBeUsed")
060public class HttpConnection implements Connection {
061    public static final String CONTENT_ENCODING = "Content-Encoding";
062    /**
063     * Many users would get caught by not setting a user-agent and therefore getting different responses on their desktop
064     * vs in jsoup, which would otherwise default to {@code Java}. So by default, use a desktop UA.
065     */
066    public static final String DEFAULT_UA =
067        "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/137.0.0.0 Safari/537.36";
068    private static final String USER_AGENT = "User-Agent";
069    public static final String CONTENT_TYPE = "Content-Type";
070    public static final String MULTIPART_FORM_DATA = "multipart/form-data";
071    public static final String FORM_URL_ENCODED = "application/x-www-form-urlencoded";
072    static final String DefaultUploadType = "application/octet-stream";
073    private static final Charset ISO_8859_1 = Charset.forName("ISO-8859-1");
074
075    private HttpConnection.Request req;
076    private Connection.@Nullable Response res;
077    @Nullable Object client; // The HttpClient for this Connection, if via the HttpClientExecutor
078    @Nullable RequestAuthenticator lastAuth; // The previous Authenticator used by this Connection, if via the HttpClientExecutor
079
080    /**
081     Create a new Connection, with the request URL specified.
082     @param url the URL to fetch from
083     @return a new Connection object
084     */
085    public static Connection connect(String url) {
086        Connection con = new HttpConnection();
087        con.url(url);
088        return con;
089    }
090
091    /**
092     Create a new Connection, with the request URL specified.
093     @param url the URL to fetch from
094     @return a new Connection object
095     */
096    public static Connection connect(URL url) {
097        Connection con = new HttpConnection();
098        con.url(url);
099        return con;
100    }
101
102    /**
103     Create a new, empty HttpConnection.
104     */
105    public HttpConnection() {
106        req = new Request();
107        req.connection = this;
108    }
109
110    /**
111     Create a new Request by deep-copying an existing Request. Note that the data and body of the original are not
112     copied. All other settings (proxy, parser, cookies, etc) are copied.
113     @param copy the request to copy
114     */
115    HttpConnection(Request copy) {
116        req = new Request(copy);
117    }
118
119    static String encodeMimeName(String val) {
120        return val.replace("\"", "%22");
121    }
122
123    @Override
124    public Connection newRequest() {
125        // copy the prototype request for the different settings, cookie manager, etc
126        return new HttpConnection(req);
127    }
128
129    /** Create a new Connection that just wraps the provided Request and Response */
130    private HttpConnection(Request req, Response res) {
131        this.req = req;
132        this.res = res;
133    }
134
135    @Override
136    public Connection url(URL url) {
137        req.url(url);
138        return this;
139    }
140
141    @Override
142    public Connection url(String url) {
143        Validate.notEmptyParam(url, "url");
144        try {
145            req.url(new URL(url));
146        } catch (MalformedURLException e) {
147            throw new IllegalArgumentException(String.format("The supplied URL, '%s', is malformed. Make sure it is an absolute URL, and starts with 'http://' or 'https://'. See https://jsoup.org/cookbook/extracting-data/working-with-urls", url), e);
148        }
149        return this;
150    }
151
152    @Override
153    public Connection proxy(@Nullable Proxy proxy) {
154        req.proxy(proxy);
155        return this;
156    }
157
158    @Override
159    public Connection proxy(String host, int port) {
160        req.proxy(host, port);
161        return this;
162    }
163
164    @Override
165    public Connection userAgent(String userAgent) {
166        Validate.notNullParam(userAgent, "userAgent");
167        req.header(USER_AGENT, userAgent);
168        return this;
169    }
170
171    @Override
172    public Connection timeout(int millis) {
173        req.timeout(millis);
174        return this;
175    }
176
177    @Override
178    public Connection maxBodySize(int bytes) {
179        req.maxBodySize(bytes);
180        return this;
181    }
182
183    @Override
184    public Connection followRedirects(boolean followRedirects) {
185        req.followRedirects(followRedirects);
186        return this;
187    }
188
189    @Override
190    public Connection referrer(String referrer) {
191        Validate.notNullParam(referrer, "referrer");
192        req.header("Referer", referrer);
193        return this;
194    }
195
196    @Override
197    public Connection method(Method method) {
198        req.method(method);
199        return this;
200    }
201
202    @Override
203    public Connection ignoreHttpErrors(boolean ignoreHttpErrors) {
204                req.ignoreHttpErrors(ignoreHttpErrors);
205                return this;
206        }
207
208    @Override
209    public Connection ignoreContentType(boolean ignoreContentType) {
210        req.ignoreContentType(ignoreContentType);
211        return this;
212    }
213
214    @Override
215    public Connection data(String key, String value) {
216        req.data(KeyVal.create(key, value));
217        return this;
218    }
219
220    @Override
221    @Deprecated
222    public Connection sslSocketFactory(SSLSocketFactory sslSocketFactory) {
223        req.sslSocketFactory(sslSocketFactory);
224        return this;
225    }
226
227    @Override
228    public Connection sslContext(SSLContext sslContext) {
229        req.sslContext(sslContext);
230        return this;
231    }
232
233    @Override
234    public Connection data(String key, String filename, InputStream inputStream) {
235        req.data(KeyVal.create(key, filename, inputStream));
236        return this;
237    }
238
239    @Override
240    public Connection data(String key, String filename, InputStream inputStream, String contentType) {
241        req.data(KeyVal.create(key, filename, inputStream).contentType(contentType));
242        return this;
243    }
244
245    @Override
246    public Connection data(Map<String, String> data) {
247        Validate.notNullParam(data, "data");
248        for (Map.Entry<String, String> entry : data.entrySet()) {
249            req.data(KeyVal.create(entry.getKey(), entry.getValue()));
250        }
251        return this;
252    }
253
254    @Override
255    public Connection data(String... keyvals) {
256        Validate.notNullParam(keyvals, "keyvals");
257        Validate.isTrue(keyvals.length %2 == 0, "Must supply an even number of key value pairs");
258        for (int i = 0; i < keyvals.length; i += 2) {
259            String key = keyvals[i];
260            String value = keyvals[i+1];
261            Validate.notEmpty(key, "Data key must not be empty");
262            Validate.notNull(value, "Data value must not be null");
263            req.data(KeyVal.create(key, value));
264        }
265        return this;
266    }
267
268    @Override
269    public Connection data(Collection<Connection.KeyVal> data) {
270        Validate.notNullParam(data, "data");
271        for (Connection.KeyVal entry: data) {
272            req.data(entry);
273        }
274        return this;
275    }
276
277    @Override
278    public Connection.@Nullable KeyVal data(String key) {
279        Validate.notEmptyParam(key, "key");
280        for (Connection.KeyVal keyVal : request().data()) {
281            if (keyVal.key().equals(key))
282                return keyVal;
283        }
284        return null;
285    }
286
287    @Override
288    public Connection requestBody(String body) {
289        req.requestBody(body);
290        return this;
291    }
292
293    @Override
294    public Connection requestBodyStream(InputStream stream) {
295        req.requestBodyStream(stream);
296        return this;
297    }
298
299    @Override
300    public Connection header(String name, String value) {
301        req.header(name, value);
302        return this;
303    }
304
305    @Override
306    public Connection headers(Map<String,String> headers) {
307        Validate.notNullParam(headers, "headers");
308        for (Map.Entry<String,String> entry : headers.entrySet()) {
309            req.header(entry.getKey(),entry.getValue());
310        }
311        return this;
312    }
313
314    @Override
315    public Connection cookie(String name, String value) {
316        req.cookie(name, value);
317        return this;
318    }
319
320    @Override
321    public Connection cookies(Map<String, String> cookies) {
322        Validate.notNullParam(cookies, "cookies");
323        for (Map.Entry<String, String> entry : cookies.entrySet()) {
324            req.cookie(entry.getKey(), entry.getValue());
325        }
326        return this;
327    }
328
329    @Override
330    public Connection cookieStore(CookieStore cookieStore) {
331        // create a new cookie manager using the new store
332        req.cookieManager = new CookieManager(cookieStore, null);
333        return this;
334    }
335
336    @Override
337    public CookieStore cookieStore() {
338        return req.cookieManager.getCookieStore();
339    }
340
341    @Override
342    public Connection parser(Parser parser) {
343        req.parser(parser);
344        return this;
345    }
346
347    @Override
348    public Document get() throws IOException {
349        req.method(Method.GET);
350        execute();
351        Validate.notNull(res);
352        return res.parse();
353    }
354
355    @Override
356    public Document post() throws IOException {
357        req.method(Method.POST);
358        execute();
359        Validate.notNull(res);
360        return res.parse();
361    }
362
363    @Override
364    public Connection.Response execute() throws IOException {
365        res = Response.execute(req);
366        return res;
367    }
368
369    @Override
370    public Connection.Request request() {
371        return req;
372    }
373
374    @Override
375    public Connection request(Connection.Request request) {
376        req = (HttpConnection.Request) request; // will throw a class-cast exception if the user has extended some but not all of Connection; that's desired
377        return this;
378    }
379
380    @Override
381    public Connection.Response response() {
382        if (res == null) {
383            throw new IllegalArgumentException("You must execute the request before getting a response.");
384        }
385        return res;
386    }
387
388    @Override
389    public Connection response(Connection.Response response) {
390        res = response;
391        return this;
392    }
393
394    @Override
395    public Connection postDataCharset(String charset) {
396        req.postDataCharset(charset);
397        return this;
398    }
399
400    @Override public Connection auth(@Nullable RequestAuthenticator authenticator) {
401        req.auth(authenticator);
402        return this;
403    }
404
405    @Override public Connection onResponseProgress(Progress<Connection.Response> handler) {
406        req.responseProgress = handler;
407        return this;
408    }
409
410    @SuppressWarnings("unchecked")
411    private static abstract class Base<T extends Connection.Base<T>> implements Connection.Base<T> {
412        private static final URL UnsetUrl; // only used if you created a new Request()
413        static {
414            try {
415                UnsetUrl = new URL("http://undefined/");
416            } catch (MalformedURLException e) {
417                throw new IllegalStateException(e);
418            }
419        }
420
421        URL url = UnsetUrl;
422        Method method = Method.GET;
423        Map<String, List<String>> headers;
424        Map<String, String> cookies;
425
426        private Base() {
427            headers = new LinkedHashMap<>();
428            cookies = new LinkedHashMap<>();
429        }
430
431        private Base(Base<T> copy) {
432            url = copy.url; // unmodifiable object
433            method = copy.method;
434            headers = new LinkedHashMap<>();
435            for (Map.Entry<String, List<String>> entry : copy.headers.entrySet()) {
436                headers.put(entry.getKey(), new ArrayList<>(entry.getValue()));
437            }
438            cookies = new LinkedHashMap<>(); cookies.putAll(copy.cookies); // just holds strings
439        }
440
441        @Override
442        public URL url() {
443            if (url == UnsetUrl)
444                throw new IllegalArgumentException("URL not set. Make sure to call #url(...) before executing the request.");
445            return url;
446        }
447
448        @Override
449        public T url(URL url) {
450            Validate.notNullParam(url, "url");
451            this.url = new UrlBuilder(url).build();
452            return (T) this;
453        }
454
455        @Override
456        public Method method() {
457            return method;
458        }
459
460        @Override
461        public T method(Method method) {
462            Validate.notNullParam(method, "method");
463            this.method = method;
464            return (T) this;
465        }
466
467        @Override @Nullable
468        public String header(String name) {
469            Validate.notNullParam(name, "name");
470            List<String> vals = getHeadersCaseInsensitive(name);
471            if (!vals.isEmpty()) {
472                // https://www.w3.org/Protocols/rfc2616/rfc2616-sec4.html#sec4.2
473                return StringUtil.join(vals, ", ");
474            }
475
476            return null;
477        }
478
479        @Override
480        public T addHeader(String name, @Nullable String value) {
481            Validate.notEmptyParam(name, "name");
482            value = value == null ? "" : value;
483
484            List<String> values = headers(name);
485            if (values.isEmpty()) {
486                values = new ArrayList<>();
487                headers.put(name, values);
488            }
489            values.add(value);
490
491            return (T) this;
492        }
493
494        @Override
495        public List<String> headers(String name) {
496            Validate.notEmptyParam(name, "name");
497            return getHeadersCaseInsensitive(name);
498        }
499
500        @Override
501        public T header(String name, String value) {
502            Validate.notEmptyParam(name, "name");
503            removeHeader(name); // ensures we don't get an "accept-encoding" and an "Accept-Encoding"
504            addHeader(name, value);
505            return (T) this;
506        }
507
508        @Override
509        public boolean hasHeader(String name) {
510            Validate.notEmptyParam(name, "name");
511            return !getHeadersCaseInsensitive(name).isEmpty();
512        }
513
514        /**
515         * Test if the request has a header with this value (case-insensitive).
516         */
517        @Override
518        public boolean hasHeaderWithValue(String name, String value) {
519            Validate.notEmpty(name);
520            Validate.notEmpty(value);
521            List<String> values = headers(name);
522            for (String candidate : values) {
523                if (value.equalsIgnoreCase(candidate))
524                    return true;
525            }
526            return false;
527        }
528
529        @Override
530        public T removeHeader(String name) {
531            Validate.notEmptyParam(name, "name");
532            Map.Entry<String, List<String>> entry = scanHeaders(name); // remove is case-insensitive too
533            if (entry != null)
534                headers.remove(entry.getKey()); // ensures correct case
535            return (T) this;
536        }
537
538        @Override
539        public Map<String, String> headers() {
540            LinkedHashMap<String, String> map = new LinkedHashMap<>(headers.size());
541            for (Map.Entry<String, List<String>> entry : headers.entrySet()) {
542                String header = entry.getKey();
543                List<String> values = entry.getValue();
544                if (!values.isEmpty())
545                    map.put(header, values.get(0));
546            }
547            return map;
548        }
549
550        @Override
551        public Map<String, List<String>> multiHeaders() {
552            return headers;
553        }
554
555        private List<String> getHeadersCaseInsensitive(String name) {
556            Validate.notNull(name);
557
558            for (Map.Entry<String, List<String>> entry : headers.entrySet()) {
559                if (name.equalsIgnoreCase(entry.getKey()))
560                    return entry.getValue();
561            }
562
563            return Collections.emptyList();
564        }
565
566        private Map.@Nullable Entry<String, List<String>> scanHeaders(String name) {
567            String lc = lowerCase(name);
568            for (Map.Entry<String, List<String>> entry : headers.entrySet()) {
569                if (lowerCase(entry.getKey()).equals(lc))
570                    return entry;
571            }
572            return null;
573        }
574
575        @Override
576        public String cookie(String name) {
577            Validate.notEmptyParam(name, "name");
578            return cookies.get(name);
579        }
580
581        @Override
582        public T cookie(String name, String value) {
583            Validate.notEmptyParam(name, "name");
584            Validate.notNullParam(value, "value");
585            cookies.put(name, value);
586            return (T) this;
587        }
588
589        @Override
590        public boolean hasCookie(String name) {
591            Validate.notEmptyParam(name, "name");
592            return cookies.containsKey(name);
593        }
594
595        @Override
596        public T removeCookie(String name) {
597            Validate.notEmptyParam(name, "name");
598            cookies.remove(name);
599            return (T) this;
600        }
601
602        @Override
603        public Map<String, String> cookies() {
604            return cookies;
605        }
606    }
607
608    public static class Request extends HttpConnection.Base<Connection.Request> implements Connection.Request {
609        static {
610            System.setProperty("sun.net.http.allowRestrictedHeaders", "true");
611            // make sure that we can send Sec-Fetch-Site headers etc.
612        }
613
614        HttpConnection connection;
615        private @Nullable Proxy proxy;
616        private int timeoutMilliseconds;
617        private int maxBodySizeBytes;
618        private boolean followRedirects;
619        private final Collection<Connection.KeyVal> data;
620        private @Nullable Object body = null; // String or InputStream
621        @Nullable String mimeBoundary;
622        private boolean ignoreHttpErrors = false;
623        private boolean ignoreContentType = false;
624        private Parser parser;
625        private boolean parserDefined = false; // called parser(...) vs initialized in ctor
626        private String postDataCharset = DataUtil.defaultCharsetName;
627        private @Nullable SSLSocketFactory sslSocketFactory;
628        @Nullable SSLContext sslContext;
629        private CookieManager cookieManager;
630        @Nullable RequestAuthenticator authenticator;
631        private @Nullable Progress<Connection.Response> responseProgress;
632
633        private final ReentrantLock executing = new ReentrantLock(); // detects and warns if same request used concurrently
634
635        Request() {
636            super();
637            timeoutMilliseconds = 30000; // 30 seconds
638            maxBodySizeBytes = 1024 * 1024 * 2; // 2MB
639            followRedirects = true;
640            data = new ArrayList<>();
641            method = Method.GET;
642            addHeader("Accept-Encoding", "gzip");
643            addHeader(USER_AGENT, DEFAULT_UA);
644            parser = Parser.htmlParser();
645            cookieManager = new CookieManager(); // creates a default InMemoryCookieStore
646        }
647
648        Request(Request copy) {
649            super(copy);
650            connection = copy.connection;
651            proxy = copy.proxy;
652            postDataCharset = copy.postDataCharset;
653            timeoutMilliseconds = copy.timeoutMilliseconds;
654            maxBodySizeBytes = copy.maxBodySizeBytes;
655            followRedirects = copy.followRedirects;
656            data = new ArrayList<>(); // data not copied
657            //body not copied
658            ignoreHttpErrors = copy.ignoreHttpErrors;
659            ignoreContentType = copy.ignoreContentType;
660            parser = copy.parser.newInstance(); // parsers and their tree-builders maintain state, so need a fresh copy
661            parserDefined = copy.parserDefined;
662            sslSocketFactory = copy.sslSocketFactory; // these are all synchronized so safe to share
663            sslContext = copy.sslContext;
664            cookieManager = copy.cookieManager;
665            authenticator = copy.authenticator;
666            responseProgress = copy.responseProgress;
667        }
668
669        @Override @Nullable
670        public Proxy proxy() {
671            return proxy;
672        }
673
674        @Override
675        public Request proxy(@Nullable Proxy proxy) {
676            this.proxy = proxy;
677            return this;
678        }
679
680        @Override
681        public Request proxy(String host, int port) {
682            this.proxy = new Proxy(Proxy.Type.HTTP, InetSocketAddress.createUnresolved(host, port));
683            return this;
684        }
685
686        @Override
687        public int timeout() {
688            return timeoutMilliseconds;
689        }
690
691        @Override
692        public Request timeout(int millis) {
693            Validate.isTrue(millis >= 0, "Timeout milliseconds must be 0 (infinite) or greater");
694            timeoutMilliseconds = millis;
695            return this;
696        }
697
698        @Override
699        public int maxBodySize() {
700            return maxBodySizeBytes;
701        }
702
703        @Override
704        public Connection.Request maxBodySize(int bytes) {
705            Validate.isTrue(bytes >= 0, "maxSize must be 0 (unlimited) or larger");
706            maxBodySizeBytes = bytes;
707            return this;
708        }
709
710        @Override
711        public boolean followRedirects() {
712            return followRedirects;
713        }
714
715        @Override
716        public Connection.Request followRedirects(boolean followRedirects) {
717            this.followRedirects = followRedirects;
718            return this;
719        }
720
721        @Override
722        public boolean ignoreHttpErrors() {
723            return ignoreHttpErrors;
724        }
725
726        @Override @Nullable
727        public SSLSocketFactory sslSocketFactory() {
728            return sslSocketFactory;
729        }
730
731        @Override
732        @Deprecated
733        public void sslSocketFactory(SSLSocketFactory sslSocketFactory) {
734            this.sslSocketFactory = sslSocketFactory;
735        }
736
737        @Override @Nullable
738        public SSLContext sslContext() {
739            return sslContext;
740        }
741
742        @Override
743        public Connection.Request sslContext(SSLContext sslContext) {
744            this.sslContext = sslContext;
745            return this;
746        }
747
748        @Override
749        public Connection.Request ignoreHttpErrors(boolean ignoreHttpErrors) {
750            this.ignoreHttpErrors = ignoreHttpErrors;
751            return this;
752        }
753
754        @Override
755        public boolean ignoreContentType() {
756            return ignoreContentType;
757        }
758
759        @Override
760        public Connection.Request ignoreContentType(boolean ignoreContentType) {
761            this.ignoreContentType = ignoreContentType;
762            return this;
763        }
764
765        @Override
766        public Request data(Connection.KeyVal keyval) {
767            Validate.notNullParam(keyval, "keyval");
768            data.add(keyval);
769            return this;
770        }
771
772        @Override
773        public Collection<Connection.KeyVal> data() {
774            return data;
775        }
776
777        @Override
778        public Connection.Request requestBody(@Nullable String body) {
779            this.body = body;
780            return this;
781        }
782
783        @Override @Nullable
784        public String requestBody() {
785            return body instanceof String ? (String) body : null;
786        }
787
788        @Override
789        public Connection.Request requestBodyStream(InputStream stream) {
790            body = stream;
791            return this;
792        }
793
794        @Override
795        public Request parser(Parser parser) {
796            this.parser = parser;
797            parserDefined = true;
798            return this;
799        }
800
801        @Override
802        public Parser parser() {
803            return parser;
804        }
805
806        @Override
807        public Connection.Request postDataCharset(String charset) {
808            Validate.notNullParam(charset, "charset");
809            if (!Charset.isSupported(charset)) throw new IllegalCharsetNameException(charset);
810            this.postDataCharset = charset;
811            return this;
812        }
813
814        @Override
815        public String postDataCharset() {
816            return postDataCharset;
817        }
818
819        CookieManager cookieManager() {
820            return cookieManager;
821        }
822
823        @Override public Connection.Request auth(@Nullable RequestAuthenticator authenticator) {
824            this.authenticator = authenticator;
825            return this;
826        }
827
828        @Override @Nullable public RequestAuthenticator auth() {
829            return authenticator;
830        }
831    }
832
833    public static class Response extends HttpConnection.Base<Connection.Response> implements Connection.Response {
834        private static final int MAX_REDIRECTS = 20;
835        private static final String LOCATION = "Location";
836        private static final String[] REDIRECT_CONTENT_HEADERS = {
837            CONTENT_ENCODING, "Content-Language", "Content-Location", CONTENT_TYPE, "Content-Length", "Digest", "Last-Modified"
838        };
839        int statusCode;
840        String statusMessage = "";
841        private @Nullable ByteBuffer byteData;
842        private @Nullable ControllableInputStream bodyStream;
843        @Nullable RequestExecutor executor;
844        private @Nullable String charset;
845        @Nullable String contentType;
846        int contentLength;
847        private boolean executed = false;
848        private boolean inputStreamRead = false;
849        private int numRedirects = 0;
850        private final HttpConnection.Request req;
851
852        /*
853         * Matches XML content types (like text/xml, image/svg+xml, application/xhtml+xml;charset=UTF8, etc)
854         */
855        private static final Pattern xmlContentTypeRxp = Pattern.compile("(\\w+)/\\w*\\+?xml.*");
856
857        /**
858         <b>Internal only! </b>Creates a dummy HttpConnection.Response, useful for testing. All actual responses
859         are created from the HttpURLConnection and fields defined.
860         */
861        Response() {
862            super();
863            statusCode = 400;
864            statusMessage = "Request not made";
865            req = new Request();
866            contentType = null;
867        }
868
869        static Response execute(HttpConnection.Request req) throws IOException {
870            return execute(req, null);
871        }
872
873        /**
874         Determines the request method for an automatic redirect, or null if the status is not automatically redirected.
875         See <a href="https://www.rfc-editor.org/rfc/rfc9110.html#section-15.4">RFC 9110, Section 15.4</a>.
876         */
877        static @Nullable Method redirectMethod(int statusCode, Method method) {
878            switch (statusCode) {
879                case 301: // Moved Permanently
880                case 302: // Found
881                    return method == Method.POST ? Method.GET : method;
882                case 303: // See Other
883                    return method == Method.HEAD ? Method.HEAD : Method.GET;
884                case 307: // Temporary Redirect
885                case 308: // Permanent Redirect
886                    return method;
887                default:
888                    return null;
889            }
890        }
891
892        /**
893         Tests if two URLs share an HTTP origin, as defined by scheme, host, and effective port.
894         See <a href="https://www.rfc-editor.org/rfc/rfc9110.html#section-4.3.1">RFC 9110, Section 4.3.1</a>.
895         */
896        static boolean sameOrigin(URL first, URL second) {
897            int firstPort = first.getPort() != -1 ? first.getPort() : first.getDefaultPort();
898            int secondPort = second.getPort() != -1 ? second.getPort() : second.getDefaultPort();
899            return first.getProtocol().equalsIgnoreCase(second.getProtocol())
900                && first.getHost().equalsIgnoreCase(second.getHost())
901                && firstPort == secondPort;
902        }
903
904        static Response execute(HttpConnection.Request req, @Nullable Response prevRes) throws IOException {
905            Validate.isTrue(req.executing.tryLock(), "Multiple threads were detected trying to execute the same request concurrently. Make sure to use Connection#newRequest() and do not share an executing request between threads.");
906            Validate.notNullParam(req, "req");
907            URL url = req.url();
908            Validate.notNull(url, "URL must be specified to connect");
909            String protocol = url.getProtocol();
910            if (!protocol.equals("http") && !protocol.equals("https"))
911                throw new MalformedURLException("Only http & https protocols supported");
912            final boolean supportsBody = req.method().hasBody();
913            final boolean hasBody = req.body != null;
914            if (!supportsBody)
915                Validate.isFalse(hasBody, "Cannot set a request body for HTTP method " + req.method());
916
917            // set up the request for execution
918            if (!req.data().isEmpty() && (!supportsBody || hasBody))
919                serialiseRequestUrl(req);
920            else if (supportsBody)
921                setOutputContentType(req);
922
923            long startTime = System.nanoTime();
924            RequestExecutor executor = RequestDispatch.get(req, prevRes);
925            Response res = null;
926            try {
927                res = executor.execute();
928
929                Method nextMethod = redirectMethod(res.statusCode, req.method());
930                if (nextMethod != null && res.hasHeader(LOCATION) && req.followRedirects()) {
931                    if (nextMethod == req.method() && (req.body instanceof InputStream || needsMultipart(req)))
932                        throw new IOException("Cannot follow redirect with a streamed request body; disable followRedirects and resend with a fresh stream");
933
934                    if (nextMethod != req.method()) {
935                        req.method(nextMethod);
936                        req.data().clear();
937                        req.requestBody(null);
938                        for (String header : REDIRECT_CONTENT_HEADERS)
939                            req.removeHeader(header);
940                    }
941
942                    String location = res.header(LOCATION);
943                    Validate.notNull(location);
944                    if (location.startsWith("http:/") && location.charAt(6) != '/') // fix broken Location: http:/temp/AAG_New/en/index.php
945                        location = location.substring(6);
946                    URL redir = StringUtil.resolve(req.url(), location);
947                    if (!sameOrigin(req.url(), redir)) {
948                        // remove sensitive headers; defense-in-depth against open redirects
949                        req.removeHeader("Authorization");
950                        req.removeHeader("Cookie");
951                        req.removeHeader("Cookie2");
952                        req.cookies().clear();
953                    }
954                    req.url(redir);
955
956                    return execute(req, res);
957                }
958                if ((res.statusCode < 200 || res.statusCode >= 400) && !req.ignoreHttpErrors())
959                        throw new HttpStatusException("HTTP error fetching URL", res.statusCode, req.url().toString());
960
961                // check that we can handle the returned content type; if not, abort before fetching it
962                String contentType = res.contentType();
963                if (contentType != null
964                        && !req.ignoreContentType()
965                        && !contentType.startsWith("text/")
966                        && !xmlContentTypeRxp.matcher(contentType).matches()
967                        )
968                    throw new UnsupportedMimeTypeException("Unhandled content type. Must be text/*, */xml, or */*+xml",
969                            contentType, req.url().toString());
970
971                // switch to the XML parser if content type is xml and not parser not explicitly set
972                if (contentType != null && xmlContentTypeRxp.matcher(contentType).matches()) {
973                    if (!req.parserDefined) req.parser(Parser.xmlParser());
974                }
975
976                res.charset = DataUtil.getCharsetFromContentType(res.contentType); // may be null, readInputStream deals with it
977                if (res.contentLength != 0 && req.method() != HEAD) { // -1 means unknown, chunked. sun throws an IO exception on 500 response with no content when trying to read body
978                    InputStream stream = executor.responseBody();
979                    if (res.hasHeaderWithValue(CONTENT_ENCODING, "gzip"))
980                        stream = new GZIPInputStream(stream);
981                    else if (res.hasHeaderWithValue(CONTENT_ENCODING, "deflate"))
982                        stream = new InflaterInputStream(stream, new Inflater(true));
983                    
984                    res.bodyStream = ControllableInputStream.wrap(
985                        stream, DefaultBufferSize, req.maxBodySize())
986                        .timeout(startTime, req.timeout());
987
988                    if (req.responseProgress != null) // set response progress listener
989                        res.bodyStream.onProgress(res.contentLength, req.responseProgress, res);
990                } else {
991                    res.byteData = DataUtil.emptyByteBuffer();
992                }
993            } catch (IOException e) {
994                if (res != null) res.safeClose(); // will be non-null if got to conn
995                throw e;
996            } finally {
997                req.executing.unlock();
998
999                // detach any thread local auth delegate
1000                if (req.authenticator != null)
1001                    AuthenticationHandler.handler.remove();
1002            }
1003
1004            res.executed = true;
1005            return res;
1006        }
1007
1008        @Override
1009        public int statusCode() {
1010            return statusCode;
1011        }
1012
1013        @Override
1014        public String statusMessage() {
1015            return statusMessage;
1016        }
1017
1018        @Override @Nullable
1019        public String charset() {
1020            return charset;
1021        }
1022
1023        @Override
1024        public Response charset(String charset) {
1025            this.charset = charset;
1026            return this;
1027        }
1028
1029        @Override @Nullable
1030        public String contentType() {
1031            return contentType;
1032        }
1033
1034        /** Called from parse() or streamParser(), validates and prepares the input stream, and aligns common settings. */
1035        private ControllableInputStream prepareParse() {
1036            Validate.isTrue(executed, "Request must be executed (with .execute(), .get(), or .post() before parsing response");
1037            ControllableInputStream stream = bodyStream;
1038            if (byteData != null) { // bytes have been read in to the buffer, parse that
1039                ByteArrayInputStream bytes = new ByteArrayInputStream(byteData.array(), 0, byteData.limit());
1040                stream = ControllableInputStream.wrap(bytes, 0); // no max
1041                inputStreamRead = false; // ok to reparse if in bytes
1042            }
1043            Validate.isFalse(inputStreamRead, "Input stream already read and parsed, cannot re-read.");
1044            Validate.notNull(stream);
1045            inputStreamRead = true;
1046            return stream;
1047        }
1048
1049        @Override public Document parse() throws IOException {
1050            ControllableInputStream stream = prepareParse();
1051            Document doc = DataUtil.parseInputStream(stream, charset, url.toExternalForm(), req.parser());
1052            doc.connection(new HttpConnection(req, this)); // because we're static, don't have the connection obj. // todo - maybe hold in the req?
1053            charset = doc.outputSettings().charset().name(); // update charset from meta-equiv, possibly
1054            safeClose();
1055            return doc;
1056        }
1057
1058        @Override public StreamParser streamParser() throws IOException {
1059            ControllableInputStream stream = prepareParse();
1060            String baseUri = url.toExternalForm();
1061            DataUtil.CharsetDoc charsetDoc = DataUtil.detectCharsetForStreamParser(stream, charset, baseUri, req.parser());
1062
1063            // set up the stream parser and rig this connection up to the parsed doc:
1064            StreamParser streamer = new StreamParser(req.parser());
1065            BufferedReader reader = new BufferedReader(new InputStreamReader(charsetDoc.input, charsetDoc.charset));
1066            streamer.parse(reader, baseUri); // initializes the parse and the document, but does not step() it
1067            streamer.document().connection(new HttpConnection(req, this));
1068            charset = charsetDoc.charset.name();
1069
1070            // we don't safeClose() as in parse(); caller must close streamParser to close InputStream stream
1071            return streamer;
1072        }
1073
1074        /**
1075         Reads the bodyStream into byteData. A no-op if already executed.
1076         */
1077        @Override
1078        public Connection.Response readFully() throws IOException {
1079            Validate.isTrue(executed, "Request must be executed (with .execute(), .get(), or .post() before getting response body");
1080            if (bodyStream != null && byteData == null) {
1081                Validate.isFalse(inputStreamRead, "Request has already been read (with .parse())");
1082                try {
1083                    byteData = DataUtil.readToByteBuffer(bodyStream, req.maxBodySize());
1084                } finally {
1085                    inputStreamRead = true;
1086                    safeClose();
1087                }
1088            }
1089            return this;
1090        }
1091
1092        /**
1093         Reads the body, but throws an UncheckedIOException if an IOException occurs.
1094         @throws UncheckedIOException if an IOException occurs
1095         */
1096        private void readByteDataUnchecked() {
1097            try {
1098                readFully();
1099            } catch (IOException e) {
1100                throw new UncheckedIOException(e);
1101            }
1102        }
1103
1104        @Override
1105        public String readBody() throws IOException {
1106            readFully();
1107            return body();
1108        }
1109
1110        @Override
1111        public String body() {
1112            readByteDataUnchecked();
1113            Validate.notNull(byteData);
1114            // charset gets set from header on execute, and from meta-equiv on parse. parse may not have happened yet
1115            String body = (charset == null ? UTF_8 : Charset.forName(charset))
1116                .decode(byteData).toString();
1117            ((Buffer)byteData).rewind(); // cast to avoid covariant return type change in jdk9
1118            return body;
1119        }
1120
1121        @Override
1122        public byte[] bodyAsBytes() {
1123            readByteDataUnchecked();
1124            Validate.notNull(byteData);
1125            Validate.isTrue(byteData.hasArray()); // we made it, so it should
1126
1127            byte[] array = byteData.array();
1128            int offset = byteData.arrayOffset();
1129            int length = byteData.limit();
1130
1131            if (offset == 0 && length == array.length) { // exact, just return it
1132                return array;
1133            } else { // trim to size
1134                byte[] exactArray = new byte[length];
1135                System.arraycopy(array, offset, exactArray, 0, length);
1136                return exactArray;
1137            }
1138        }
1139
1140        @Override
1141        @Deprecated
1142        public Connection.Response bufferUp() {
1143            readByteDataUnchecked();
1144            return this;
1145        }
1146
1147        @Override
1148        public BufferedInputStream bodyStream() {
1149            Validate.isTrue(executed, "Request must be executed (with .execute(), .get(), or .post() before getting response body");
1150
1151            // if we have read to bytes (via readFully), return those as a stream.
1152            if (byteData != null) {
1153                return new BufferedInputStream(
1154                    new ByteArrayInputStream(byteData.array(), 0, byteData.limit()),
1155                    DefaultBufferSize);
1156            }
1157
1158            Validate.isFalse(inputStreamRead, "Request has already been read");
1159            Validate.notNull(bodyStream);
1160            inputStreamRead = true;
1161            return bodyStream.inputStream();
1162        }
1163
1164        /**
1165         * Call on completion of stream read, to close the body (or error) stream. The connection.disconnect allows
1166         * keep-alives to work (as the underlying connection is actually held open, despite the name).
1167         */
1168        private void safeClose() {
1169            if (bodyStream != null) {
1170                try {
1171                    bodyStream.close();
1172                } catch (IOException e) {
1173                    // no-op
1174                } finally {
1175                    bodyStream = null;
1176                }
1177            }
1178
1179            if (executor != null) executor.safeClose(); // disconnect
1180        }
1181
1182        Response(HttpConnection.Request request) {
1183            this.req = request;
1184        }
1185
1186        // set up url, method, header, cookies
1187        void prepareResponse(Map<String, List<String>> resHeaders, HttpConnection.@Nullable Response previousResponse) throws IOException {
1188            processResponseHeaders(resHeaders); // includes cookie key/val read during header scan
1189            CookieUtil.storeCookies(req, this, url, resHeaders); // add set cookies to cookie store
1190
1191            if (previousResponse != null) { // was redirected
1192                // map previous response cookies into this response cookies() object
1193                for (Map.Entry<String, String> prevCookie : previousResponse.cookies().entrySet()) {
1194                    if (!hasCookie(prevCookie.getKey()))
1195                        cookie(prevCookie.getKey(), prevCookie.getValue());
1196                }
1197                previousResponse.safeClose();
1198
1199                // enforce too many redirects:
1200                numRedirects = previousResponse.numRedirects + 1;
1201                if (numRedirects >= MAX_REDIRECTS)
1202                    throw new IOException(String.format("Too many redirects occurred trying to load URL %s", previousResponse.url()));
1203            }
1204        }
1205
1206        void processResponseHeaders(Map<String, List<String>> resHeaders) {
1207            for (Map.Entry<String, List<String>> entry : resHeaders.entrySet()) {
1208                String name = entry.getKey();
1209                if (name == null)
1210                    continue; // http/1.1 line
1211
1212                List<String> values = entry.getValue();
1213                for (String value : values) {
1214                    addHeader(name, fixHeaderEncoding(value));
1215                }
1216            }
1217        }
1218
1219        /**
1220         Servers may encode response headers in UTF-8 instead of RFC defined 8859. The JVM decodes the headers (before we see them) as 8859, which can lead to mojibake data.
1221         <p>This method attempts to detect that and re-decode the string as UTF-8.</p>
1222         <p>However on Android, the headers will be decoded as UTF8, so we can detect and pass those directly.</p>
1223         * @param val a header value string that may have been incorrectly decoded as 8859.
1224         * @return a potentially re-decoded string.
1225         */
1226        @Nullable
1227        static String fixHeaderEncoding(@Nullable String val) {
1228            if (val == null) return val;
1229            // If we can't encode the string as 8859, then it couldn't have been decoded as 8859
1230            if (!StandardCharsets.ISO_8859_1.newEncoder().canEncode(val))
1231                return val;
1232            byte[] bytes = val.getBytes(ISO_8859_1);
1233            if (looksLikeUtf8(bytes))
1234                return new String(bytes, UTF_8);
1235            else
1236                return val;
1237        }
1238
1239        private static boolean looksLikeUtf8(byte[] input) {
1240            int i = 0;
1241            // BOM:
1242            if (input.length >= 3
1243                && (input[0] & 0xFF) == 0xEF
1244                && (input[1] & 0xFF) == 0xBB
1245                && (input[2] & 0xFF) == 0xBF) {
1246                i = 3;
1247            }
1248
1249            int end;
1250            boolean foundNonAscii = false;
1251            for (int j = input.length; i < j; ++i) {
1252                int o = input[i];
1253                if ((o & 0x80) == 0) {
1254                    continue; // ASCII
1255                }
1256                foundNonAscii = true;
1257
1258                // UTF-8 leading:
1259                if ((o & 0xE0) == 0xC0) {
1260                    end = i + 1;
1261                } else if ((o & 0xF0) == 0xE0) {
1262                    end = i + 2;
1263                } else if ((o & 0xF8) == 0xF0) {
1264                    end = i + 3;
1265                } else {
1266                    return false;
1267                }
1268
1269                if (end >= input.length)
1270                    return false;
1271
1272                while (i < end) {
1273                    i++;
1274                    o = input[i];
1275                    if ((o & 0xC0) != 0x80) {
1276                        return false;
1277                    }
1278                }
1279            }
1280            return foundNonAscii;
1281        }
1282
1283        private static void setOutputContentType(final HttpConnection.Request req) {
1284            final String contentType = req.header(CONTENT_TYPE);
1285            String bound = contentType != null && contentType.contains(MULTIPART_FORM_DATA) ? req.mimeBoundary : null;
1286            if (contentType != null) {
1287                // no-op; don't add content type as already set (e.g. for requestBody())
1288                // todo - if content type already set, we could add charset
1289
1290                // if user has set content type to multipart/form-data, auto add boundary.
1291                if(contentType.contains(MULTIPART_FORM_DATA) && !contentType.contains("boundary")) {
1292                    bound = DataUtil.mimeBoundary();
1293                    req.header(CONTENT_TYPE, MULTIPART_FORM_DATA + "; boundary=" + bound);
1294                }
1295
1296            }
1297            else if (needsMultipart(req)) {
1298                bound = DataUtil.mimeBoundary();
1299                req.header(CONTENT_TYPE, MULTIPART_FORM_DATA + "; boundary=" + bound);
1300            } else {
1301                req.header(CONTENT_TYPE, FORM_URL_ENCODED + "; charset=" + req.postDataCharset());
1302            }
1303            req.mimeBoundary = bound;
1304        }
1305
1306        static void writePost(final HttpConnection.Request req, final OutputStream outputStream) throws IOException {
1307            try (OutputStreamWriter osw = new OutputStreamWriter(outputStream, req.postDataCharset());
1308                 BufferedWriter w = new BufferedWriter(osw)) {
1309                implWritePost(req, w, outputStream);
1310            }
1311        }
1312
1313        private static void implWritePost(final HttpConnection.Request req, final BufferedWriter w, final OutputStream outputStream) throws IOException {
1314            final Collection<Connection.KeyVal> data = req.data();
1315            final String boundary = req.mimeBoundary;
1316
1317            if (boundary != null) { // a multipart post
1318                for (Connection.KeyVal keyVal : data) {
1319                    w.write("--");
1320                    w.write(boundary);
1321                    w.write("\r\n");
1322                    w.write("Content-Disposition: form-data; name=\"");
1323                    w.write(encodeMimeName(keyVal.key())); // encodes " to %22
1324                    w.write("\"");
1325                    final InputStream input = keyVal.inputStream();
1326                    if (input != null) {
1327                        w.write("; filename=\"");
1328                        w.write(encodeMimeName(keyVal.value()));
1329                        w.write("\"\r\nContent-Type: ");
1330                        String contentType = keyVal.contentType();
1331                        w.write(contentType != null ? contentType : DefaultUploadType);
1332                        w.write("\r\n\r\n");
1333                        w.flush();
1334                        DataUtil.crossStreams(input, outputStream);
1335                        outputStream.flush();
1336                    } else {
1337                        w.write("\r\n\r\n");
1338                        w.write(keyVal.value());
1339                    }
1340                    w.write("\r\n");
1341                }
1342                w.write("--");
1343                w.write(boundary);
1344                w.write("--");
1345            } else if (req.body != null) { // a single body (bytes or plain text);  data will be in query string
1346                if (req.body instanceof String) {
1347                    w.write((String) req.body);
1348                } else if (req.body instanceof InputStream) {
1349                    DataUtil.crossStreams((InputStream) req.body, outputStream);
1350                    outputStream.flush();
1351                } else {
1352                    throw new IllegalStateException();
1353                }
1354            } else { // regular form data (application/x-www-form-urlencoded)
1355                boolean first = true;
1356                for (Connection.KeyVal keyVal : data) {
1357                    if (!first) w.append('&');
1358                    else first = false;
1359
1360                    w.write(URLEncoder.encode(keyVal.key(), req.postDataCharset()));
1361                    w.write('=');
1362                    w.write(URLEncoder.encode(keyVal.value(), req.postDataCharset()));
1363                }
1364            }
1365        }
1366
1367        // for get url reqs, serialise the data map into the url
1368        private static void serialiseRequestUrl(Connection.Request req) throws IOException {
1369            UrlBuilder in = new UrlBuilder(req.url());
1370
1371            for (Connection.KeyVal keyVal : req.data()) {
1372                Validate.isFalse(keyVal.hasInputStream(), "InputStream data not supported in URL query string.");
1373                in.appendKeyVal(keyVal);
1374            }
1375            req.url(in.build());
1376            req.data().clear(); // moved into url as get params
1377        }
1378    }
1379
1380    private static boolean needsMultipart(Connection.Request req) {
1381        // multipart mode, for files. add the header if we see something with an inputstream, and return a non-null boundary
1382        for (Connection.KeyVal keyVal : req.data()) {
1383            if (keyVal.hasInputStream())
1384                return true;
1385        }
1386        return false;
1387    }
1388
1389    public static class KeyVal implements Connection.KeyVal {
1390        private String key;
1391        private String value;
1392        private @Nullable InputStream stream;
1393        private @Nullable String contentType;
1394
1395        public static KeyVal create(String key, String value) {
1396            return new KeyVal(key, value);
1397        }
1398
1399        public static KeyVal create(String key, String filename, InputStream stream) {
1400            return new KeyVal(key, filename)
1401                .inputStream(stream);
1402        }
1403
1404        private KeyVal(String key, String value) {
1405            Validate.notEmptyParam(key, "key");
1406            Validate.notNullParam(value, "value");
1407            this.key = key;
1408            this.value = value;
1409        }
1410
1411        @Override
1412        public KeyVal key(String key) {
1413            Validate.notEmptyParam(key, "key");
1414            this.key = key;
1415            return this;
1416        }
1417
1418        @Override
1419        public String key() {
1420            return key;
1421        }
1422
1423        @Override
1424        public KeyVal value(String value) {
1425            Validate.notNullParam(value, "value");
1426            this.value = value;
1427            return this;
1428        }
1429
1430        @Override
1431        public String value() {
1432            return value;
1433        }
1434
1435        @Override
1436        public KeyVal inputStream(InputStream inputStream) {
1437            Validate.notNullParam(inputStream, "inputStream");
1438            this.stream = inputStream;
1439            return this;
1440        }
1441
1442        @Override @Nullable
1443        public InputStream inputStream() {
1444            return stream;
1445        }
1446
1447        @Override
1448        public boolean hasInputStream() {
1449            return stream != null;
1450        }
1451
1452        @Override
1453        public Connection.KeyVal contentType(String contentType) {
1454            Validate.notEmpty(contentType);
1455            this.contentType = contentType;
1456            return this;
1457        }
1458
1459        @Override @Nullable
1460        public String contentType() {
1461            return contentType;
1462        }
1463
1464        @Override
1465        public String toString() {
1466            return key + "=" + value;
1467        }
1468    }
1469}