diff options
4 files changed, 31 insertions, 11 deletions
diff --git a/main/src/cgeo/geocaching/connector/gc/GCConstants.java b/main/src/cgeo/geocaching/connector/gc/GCConstants.java index e9aa4e2..1596a46 100644 --- a/main/src/cgeo/geocaching/connector/gc/GCConstants.java +++ b/main/src/cgeo/geocaching/connector/gc/GCConstants.java @@ -36,7 +36,7 @@ public final class GCConstants { public final static Pattern PATTERN_LATLON = Pattern.compile("<span id=\"uxLatLon\"[^>]*>(.*?)</span>"); public final static Pattern PATTERN_LATLON_ORIG = Pattern.compile("\\{\"isUserDefined\":true[^}]+?\"oldLatLngDisplay\":\"([^\"]+)\"\\}"); public final static Pattern PATTERN_LOCATION = Pattern.compile(Pattern.quote("<span id=\"ctl00_ContentBody_Location\">In ") + "(?:<a href=[^>]*>)?(.*?)<"); - public final static Pattern PATTERN_PERSONALNOTE = Pattern.compile("<p id=\"cache_note\"[^>]*>(.*?)</p>"); + public final static Pattern PATTERN_PERSONALNOTE = Pattern.compile("<p id=\"cache_note\"[^>]*>(.*?)</p>", Pattern.DOTALL); public final static Pattern PATTERN_NAME = Pattern.compile("<span id=\"ctl00_ContentBody_CacheName\">(.*?)</span>"); public final static Pattern PATTERN_DIFFICULTY = Pattern.compile("<span id=\"ctl00_ContentBody_uxLegendScale\"[^>]*>[^<]*<img src=\"[^\"]*/images/stars/stars([0-9_]+)\\.gif\""); public final static Pattern PATTERN_TERRAIN = Pattern.compile("<span id=\"ctl00_ContentBody_Localize[\\d]+\"[^>]*>[^<]*<img src=\"[^\"]*/images/stars/stars([0-9_]+)\\.gif\""); diff --git a/main/src/cgeo/geocaching/connector/gc/GCParser.java b/main/src/cgeo/geocaching/connector/gc/GCParser.java index 39a8891..cebb3ca 100644 --- a/main/src/cgeo/geocaching/connector/gc/GCParser.java +++ b/main/src/cgeo/geocaching/connector/gc/GCParser.java @@ -356,32 +356,41 @@ public abstract class GCParser { return searchResult; } - static SearchResult parseCacheFromText(final String page, final CancellableHandler handler) { + static SearchResult parseCacheFromText(final String pageIn, final CancellableHandler handler) { CancellableHandler.sendLoadProgressDetail(handler, R.string.cache_dialog_loading_details_status_details); - if (StringUtils.isBlank(page)) { + if (StringUtils.isBlank(pageIn)) { Log.e("GCParser.parseCache: No page given"); return null; } final SearchResult searchResult = new SearchResult(); - if (page.contains(GCConstants.STRING_UNPUBLISHED_OTHER) || page.contains(GCConstants.STRING_UNPUBLISHED_OWNER) || page.contains(GCConstants.STRING_UNPUBLISHED_FROM_SEARCH)) { + if (pageIn.contains(GCConstants.STRING_UNPUBLISHED_OTHER) || pageIn.contains(GCConstants.STRING_UNPUBLISHED_OWNER) || pageIn.contains(GCConstants.STRING_UNPUBLISHED_FROM_SEARCH)) { searchResult.setError(StatusCode.UNPUBLISHED_CACHE); return searchResult; } - if (page.contains(GCConstants.STRING_PREMIUMONLY_1) || page.contains(GCConstants.STRING_PREMIUMONLY_2)) { + if (pageIn.contains(GCConstants.STRING_PREMIUMONLY_1) || pageIn.contains(GCConstants.STRING_PREMIUMONLY_2)) { searchResult.setError(StatusCode.PREMIUM_ONLY); return searchResult; } - final String cacheName = Html.fromHtml(TextUtils.getMatch(page, GCConstants.PATTERN_NAME, true, "")).toString(); + final String cacheName = Html.fromHtml(TextUtils.getMatch(pageIn, GCConstants.PATTERN_NAME, true, "")).toString(); if (GCConstants.STRING_UNKNOWN_ERROR.equalsIgnoreCase(cacheName)) { searchResult.setError(StatusCode.UNKNOWN_ERROR); return searchResult; } + // first handle the content with line breaks, then trim everything for easier matching and reduced memory consumption in parsed fields + String personalNoteWithLineBreaks = ""; + MatcherWrapper matcher = new MatcherWrapper(GCConstants.PATTERN_PERSONALNOTE, pageIn); + if (matcher.find()) { + personalNoteWithLineBreaks = matcher.group(1).trim(); + } + + final String page = TextUtils.replaceWhitespace(pageIn); + final Geocache cache = new Geocache(); cache.setDisabled(page.contains(GCConstants.STRING_DISABLED)); @@ -518,7 +527,7 @@ public abstract class GCParser { cache.checkFields(); // cache personal note - cache.setPersonalNote(TextUtils.getMatch(page, GCConstants.PATTERN_PERSONALNOTE, true, cache.getPersonalNote())); + cache.setPersonalNote(personalNoteWithLineBreaks); // cache short description cache.setShortDescription(TextUtils.getMatch(page, GCConstants.PATTERN_SHORTDESC, true, "")); diff --git a/main/src/cgeo/geocaching/connector/gc/Login.java b/main/src/cgeo/geocaching/connector/gc/Login.java index 3146712..1f5714e 100644 --- a/main/src/cgeo/geocaching/connector/gc/Login.java +++ b/main/src/cgeo/geocaching/connector/gc/Login.java @@ -458,20 +458,31 @@ public abstract class Login { * @return */ public static String getRequestLogged(final String uri, final Parameters params) { - final String data = Network.getResponseData(Network.getRequest(uri, params)); + final String data = Network.getResponseData(Network.getRequest(uri, params), canRemoveWhitespace(uri)); if (getLoginStatus(data)) { return data; } if (login() == StatusCode.NO_ERROR) { - return Network.getResponseData(Network.getRequest(uri, params)); + return Network.getResponseData(Network.getRequest(uri, params), canRemoveWhitespace(uri)); } Log.w("Working as guest."); return data; } + /** + * Unfortunately the cache details page contains user generated whitespace in the personal note, therefore we cannot + * remove the white space from cache details pages. + * + * @param uri + * @return + */ + private static boolean canRemoveWhitespace(final String uri) { + return !StringUtils.contains(uri, "cache_details"); + } + /** Get user session & session token from the Live Map. Needed for following requests */ public static String[] getMapTokens() { final HttpResponse response = Network.getRequest(GCConstants.URL_LIVE_MAP); diff --git a/main/src/cgeo/geocaching/utils/TextUtils.java b/main/src/cgeo/geocaching/utils/TextUtils.java index 68ac595..c9d4958 100644 --- a/main/src/cgeo/geocaching/utils/TextUtils.java +++ b/main/src/cgeo/geocaching/utils/TextUtils.java @@ -108,13 +108,13 @@ public final class TextUtils { } /** - * Replaces every \n, \r and \t with a single space. Afterwards multiples spaces + * Replaces every \n, \r and \t with a single space. Afterwards multiple spaces * are merged into a single space. Finally leading spaces are deleted. * * This method must be fast, but may not lead to the shortest replacement String. * * You are only allowed to change this code if you can prove it became faster on a device. - * see cgeo.geocaching.test.WhiteSpaceTest#replaceWhitespaceManually in the test project + * see cgeo.geocaching.test.WhiteSpaceTest#replaceWhitespaceManually in the test project. * * @param data * complete HTML page |