Epub阅读器0.0.1

2026-05-21 19:40:51 +08:00
commit daa36d8fe7
559 changed files with 106266 additions and 0 deletions
@@ -0,0 +1,607 @@
+//
+//  WREpubParser.m
+//  WeRead (微信读书)
+//  Reverse-engineered implementation reconstruction
+//
+//  Detailed pseudo-code based on binary analysis, ivar types, known methods,
+//  and contextual knowledge of EPUB file format handling.
+//
+
+#import "WREpubParser.h"
+#import "WRBook.h"
+#import "WHAlbumInfo.h"
+
+// ---------------------------------------------------------------------------
+// Constants
+// ---------------------------------------------------------------------------
+NSString *const WREpubParserErrorDomain = @"WREpubParserErrorDomain";
+
+static NSString *const kContainerXMLPath = @"META-INF/container.xml";
+static NSString *const kOPFMIMEType      = @"application/oebps-package+xml";
+static NSString *const kNCXMIMEType      = @"application/x-dtbncx+xml";
+
+#pragma mark - Private Interface
+
+@interface WREpubParser ()
+
+@property (nonatomic, copy,   readwrite) NSString         *epubFilePath;
+@property (nonatomic, copy,   readwrite) NSString         *baseDirectory;
+@property (nonatomic, strong, readwrite, nullable) NSError *lastError;
+@property (nonatomic, strong, readwrite) NSArray<NSDictionary *> *chapters;
+@property (nonatomic, strong, readwrite) NSDictionary<NSString *, NSString *> *resourceMap;
+@property (nonatomic, strong, readwrite) NSArray<NSString *> *spineItemIDs;
+
+@end
+
+#pragma mark - Implementation
+
+@implementation WREpubParser
+{
+    // Ivars confirmed from binary analysis:
+    NSString         *_epubFilePath;
+    NSString         *_baseDirectory;
+    NSError          *_lastError;
+    UIViewController *_epubController;     // weak, assigned from delegate
+    WRBook           *_book;
+    WHAlbumInfo      *_albumInfo;
+
+    // Internal caches (not exported in header but inferred):
+    NSMutableDictionary<NSString *, NSString *> *_manifestMap;   // id -> href
+    NSMutableDictionary<NSString *, NSString *> *_mediaTypeMap;  // id -> media-type
+    NSMutableArray<NSString *>                  *_spineRefs;     // idref list
+}
+
+#pragma mark - Lifecycle
+
+- (instancetype)initWithFilePath:(NSString *)path
+                            book:(nullable WRBook *)book
+{
+    self = [super init];
+    if (self) {
+        _epubFilePath  = [path copy];
+        _book          = book;
+        _manifestMap   = [NSMutableDictionary dictionary];
+        _mediaTypeMap  = [NSMutableDictionary dictionary];
+        _spineRefs     = [NSMutableArray array];
+        _chapters      = @[];
+        _resourceMap   = @{};
+        _spineItemIDs  = @[];
+    }
+    return self;
+}
+
+#pragma mark - Public: Parsing
+
+- (BOOL)parse:(NSError *_Nullable *_Nullable)error
+{
+    // 1. Locate the OPF path from container.xml
+    NSError *containerError = nil;
+    NSString *opfRelPath = [self parseContainerXML:&containerError];
+    if (!opfRelPath) {
+        [self _setError:error
+                   code:WREpubParserErrorContainerParseFail
+            description:@"Failed to parse container.xml"
+         underlyingError:containerError];
+        return NO;
+    }
+
+    // 2. Parse the OPF to populate manifest, spine, and metadata
+    NSError *opfError = nil;
+    if (![self parseOPFAtRelativePath:opfRelPath error:&opfError]) {
+        [self _setError:error
+                   code:WREpubParserErrorOPFParseFail
+            description:@"Failed to parse content.opf"
+         underlyingError:opfError];
+        return NO;
+    }
+
+    // 3. Optionally parse NCX for table of contents
+    NSError *ncxError = nil;
+    [self parseNCX:&ncxError];
+    // NCX failure is non-fatal; log but continue
+    if (ncxError) {
+        NSLog(@"[WREpubParser] NCX parse warning: %@", ncxError);
+    }
+
+    // 4. Build the chapter list from the spine
+    [self _buildChapterList];
+
+    // 5. Build the resource map from the manifest
+    [self _buildResourceMap];
+
+    if (self.chapters.count == 0) {
+        [self _setError:error
+                   code:WREpubParserErrorSpineEmpty
+            description:@"Spine contains no items"
+         underlyingError:nil];
+        return NO;
+    }
+
+    return YES;
+}
+
+// ---------------------------------------------------------------------------
+// Parse META-INF/container.xml
+// ---------------------------------------------------------------------------
+- (nullable NSString *)parseContainerXML:(NSError *_Nullable *_Nullable)error
+{
+    NSString *containerPath =
+        [self.epubFilePath stringByAppendingPathComponent:kContainerXMLPath];
+
+    // container.xml is always plain text (unencrypted)
+    NSData *data = [NSData dataWithContentsOfFile:containerPath options:0 error:error];
+    if (!data) return nil;
+
+    // Use NSXMLParser to extract the rootfile full-path
+    //
+    // Expected structure:
+    //   <container>
+    //     <rootfiles>
+    //       <rootfile full-path="OEBPS/content.opf" media-type="..."/>
+    //     </rootfiles>
+    //   </container>
+
+    NSXMLParser *parser = [[NSXMLParser alloc] initWithData:data];
+    WREpubContainerParserDelegate *delegate =
+        [[WREpubContainerParserDelegate alloc] init];
+    parser.delegate = delegate;
+
+    if (![parser parse]) {
+        if (error) *error = parser.parserError;
+        return nil;
+    }
+
+    return delegate.rootFilePath;
+}
+
+// ---------------------------------------------------------------------------
+// Parse the OPF (content.opf) file
+// ---------------------------------------------------------------------------
+- (BOOL)parseOPFAtRelativePath:(NSString *)opfRelPath
+                         error:(NSError *_Nullable *_Nullable)error
+{
+    NSString *opfFullPath =
+        [self.epubFilePath stringByAppendingPathComponent:opfRelPath];
+
+    // Set the base directory for resolving relative paths within the OPF
+    _baseDirectory = [opfFullPath stringByDeletingLastPathComponent];
+
+    NSData *data = [NSData dataWithContentsOfFile:opfFullPath options:0 error:error];
+    if (!data) return NO;
+
+    // Use NSXMLParser to walk the OPF XML
+    //
+    // Sections to parse:
+    //   <manifest>  -> populate _manifestMap and _mediaTypeMap
+    //   <spine>     -> populate _spineRefs (ordered idref list)
+    //   <metadata>  -> extract title, identifier, etc. for WRBook
+
+    WREpubOPFParserDelegate *opfDelegate =
+        [[WREpubOPFParserDelegate alloc] initWithBaseDirectory:_baseDirectory];
+    NSXMLParser *parser = [[NSXMLParser alloc] initWithData:data];
+    parser.delegate = opfDelegate;
+
+    if (![parser parse]) {
+        if (error) *error = parser.parserError;
+        return NO;
+    }
+
+    // Transfer parsed data
+    [_manifestMap  setDictionary:opfDelegate.manifestItems];
+    [_mediaTypeMap setDictionary:opfDelegate.mediaTypes];
+    [_spineRefs    setArray:opfDelegate.spineItemRefs];
+
+    // Extract metadata into _book if available
+    if (opfDelegate.bookTitle) {
+        _book.title = opfDelegate.bookTitle;
+    }
+
+    return YES;
+}
+
+// ---------------------------------------------------------------------------
+// Parse toc.ncx for table of contents
+// ---------------------------------------------------------------------------
+- (nullable NSArray *)parseNCX:(NSError *_Nullable *_Nullable)error
+{
+    // Find the NCX item in the manifest (media-type = application/x-dtbncx+xml)
+    NSString *ncxID = nil;
+    for (NSString *itemID in _mediaTypeMap) {
+        if ([_mediaTypeMap[itemID] isEqualToString:kNCXMIMEType]) {
+            ncxID = itemID;
+            break;
+        }
+    }
+
+    if (!ncxID) {
+        // Some EPUBs use nav.xhtml instead of NCX (EPUB3)
+        // Try finding nav document
+        return nil;
+    }
+
+    NSString *ncxRelPath = _manifestMap[ncxID];
+    if (!ncxRelPath) return nil;
+
+    NSString *ncxFullPath =
+        [_baseDirectory stringByAppendingPathComponent:ncxRelPath];
+
+    NSData *data = [NSData dataWithContentsOfFile:ncxFullPath options:0 error:error];
+    if (!data) return nil;
+
+    // Parse NCX XML:
+    //   <ncx>
+    //     <navMap>
+    //       <navPoint id="..." playOrder="1">
+    //         <navLabel><text>Chapter 1</text></navLabel>
+    //         <content src="chapter1.xhtml"/>
+    //         <navPoint> ... nested ... </navPoint>
+    //       </navPoint>
+    //     </navMap>
+    //   </ncx>
+
+    WREpubNCXParserDelegate *ncxDelegate = [[WREpubNCXParserDelegate alloc] init];
+    NSXMLParser *parser = [[NSXMLParser alloc] initWithData:data];
+    parser.delegate = ncxDelegate;
+
+    if (![parser parse]) {
+        if (error) *error = parser.parserError;
+        return nil;
+    }
+
+    return ncxDelegate.tocEntries;
+}
+
+// ---------------------------------------------------------------------------
+// Return chapter content (XHTML) for a given index
+// ---------------------------------------------------------------------------
+- (nullable NSString *)contentForChapterAtIndex:(NSUInteger)index
+                                          error:(NSError *_Nullable *_Nullable)error
+{
+    if (index >= self.chapters.count) return nil;
+
+    NSDictionary *chapterInfo = self.chapters[index];
+    NSString *href = chapterInfo[@"href"];
+    if (!href) return nil;
+
+    NSString *fullPath =
+        [_baseDirectory stringByAppendingPathComponent:href];
+
+    // If the file is encrypted, it must be decrypted first via WREncryptedFileManager
+    NSData *data = [NSData dataWithContentsOfFile:fullPath options:0 error:error];
+    if (!data) return nil;
+
+    return [[NSString alloc] initWithData:data encoding:NSUTF8StringEncoding];
+}
+
+// ---------------------------------------------------------------------------
+// Resolve a relative resource path
+// ---------------------------------------------------------------------------
+- (nullable NSString *)absolutePathForResource:(NSString *)relativePath
+{
+    if (!relativePath) return nil;
+
+    // Try the resource map first
+    NSString *mapped = self.resourceMap[relativePath];
+    if (mapped) return mapped;
+
+    // Fallback: resolve relative to the base directory
+    return [_baseDirectory stringByAppendingPathComponent:relativePath];
+}
+
+// ---------------------------------------------------------------------------
+// Delegate callback
+// ---------------------------------------------------------------------------
+- (void)notifyDelegateOfError:(NSError *)error
+{
+    _lastError = error;
+
+    // The delegate protocol method uses epubController:didFailWithError:
+    // The "controller" here is the UIViewController that owns the reader.
+    if ([self.delegate respondsToSelector:@selector(epubController:didFailWithError:)]) {
+        [self.delegate epubController:_epubController didFailWithError:error];
+    }
+}
+
+#pragma mark - Private Helpers
+
+/// Build the chapters array from spine references + manifest.
+- (void)_buildChapterList
+{
+    NSMutableArray *chapters = [NSMutableArray arrayWithCapacity:_spineRefs.count];
+
+    for (NSString *idref in _spineRefs) {
+        NSString *href      = _manifestMap[idref];
+        NSString *mediaType = _mediaTypeMap[idref];
+
+        if (!href) continue;
+
+        NSMutableDictionary *chapterInfo = [NSMutableDictionary dictionary];
+        chapterInfo[@"id"]        = idref;
+        chapterInfo[@"href"]      = href;
+        chapterInfo[@"mediaType"] = mediaType ?: @"application/xhtml+xml";
+        chapterInfo[@"fullPath"]  =
+            [_baseDirectory stringByAppendingPathComponent:href];
+
+        [chapters addObject:[chapterInfo copy]];
+    }
+
+    _chapters = [chapters copy];
+}
+
+/// Build a flat resource map for images, CSS, fonts, etc.
+- (void)_buildResourceMap
+{
+    NSMutableDictionary *map = [NSMutableDictionary dictionary];
+
+    for (NSString *itemID in _manifestMap) {
+        NSString *href = _manifestMap[itemID];
+        if (!href) continue;
+
+        NSString *absPath =
+            [_baseDirectory stringByAppendingPathComponent:href];
+        map[href] = absPath;
+
+        // Also index by filename for convenience
+        NSString *filename = [href lastPathComponent];
+        if (filename) {
+            map[filename] = absPath;
+        }
+    }
+
+    _resourceMap = [map copy];
+}
+
+/// Helper to set the error pointer and store lastError.
+- (void)_setError:(NSError *_Nullable *_Nullable)outError
+             code:(WREpubParserErrorCode)code
+      description:(NSString *)description
+   underlyingError:(nullable NSError *)underlyingError
+{
+    NSMutableDictionary *userInfo = [NSMutableDictionary dictionary];
+    userInfo[NSLocalizedDescriptionKey] = description;
+    if (underlyingError) {
+        userInfo[NSUnderlyingErrorKey] = underlyingError;
+    }
+
+    NSError *err = [NSError errorWithDomain:WREpubParserErrorDomain
+                                       code:code
+                                   userInfo:userInfo];
+    _lastError = err;
+    if (outError) *outError = err;
+}
+
+@end
+
+// ===========================================================================
+// Internal XML Parser Delegates (file-private)
+// ===========================================================================
+
+#pragma mark - Container Parser Delegate
+
+/// Parses META-INF/container.xml to extract the rootfile path.
+@interface WREpubContainerParserDelegate : NSObject <NSXMLParserDelegate>
+@property (nonatomic, copy, nullable) NSString *rootFilePath;
+@end
+
+@implementation WREpubContainerParserDelegate
+{
+    BOOL _insideRootfile;
+}
+
+- (void)parser:(NSXMLParser *)parser
+didStartElement:(NSString *)elementName
+  namespaceURI:(NSString *)namespaceURI
+ qualifiedName:(NSString *)qName
+    attributes:(NSDictionary<NSString *,NSString *> *)attributeDict
+{
+    if ([elementName isEqualToString:@"rootfile"]) {
+        _rootFilePath = attributeDict[@"full-path"];
+        _insideRootfile = YES;
+    }
+}
+
+- (void)parser:(NSXMLParser *)parser
+ didEndElement:(NSString *)elementName
+  namespaceURI:(NSString *)namespaceURI
+ qualifiedName:(NSString *)qName
+{
+    if ([elementName isEqualToString:@"rootfile"]) {
+        _insideRootfile = NO;
+    }
+}
+
+@end
+
+#pragma mark - OPF Parser Delegate
+
+/// Parses the OPF file: manifest, spine, and metadata sections.
+@interface WREpubOPFParserDelegate : NSObject <NSXMLParserDelegate>
+
+@property (nonatomic, strong) NSMutableDictionary<NSString *, NSString *> *manifestItems;
+@property (nonatomic, strong) NSMutableDictionary<NSString *, NSString *> *mediaTypes;
+@property (nonatomic, strong) NSMutableArray<NSString *> *spineItemRefs;
+@property (nonatomic, copy,   nullable) NSString *bookTitle;
+
+- (instancetype)initWithBaseDirectory:(NSString *)baseDir;
+
+@end
+
+@implementation WREpubOPFParserDelegate
+{
+    NSString      *_baseDirectory;
+    BOOL           _inMetadata;
+    BOOL           _inManifest;
+    BOOL           _inSpine;
+    NSMutableString *_currentText;
+}
+
+- (instancetype)initWithBaseDirectory:(NSString *)baseDir
+{
+    self = [super init];
+    if (self) {
+        _baseDirectory  = baseDir;
+        _manifestItems  = [NSMutableDictionary dictionary];
+        _mediaTypes     = [NSMutableDictionary dictionary];
+        _spineItemRefs  = [NSMutableArray array];
+    }
+    return self;
+}
+
+- (void)parser:(NSXMLParser *)parser
+didStartElement:(NSString *)elementName
+  namespaceURI:(NSString *)namespaceURI
+ qualifiedName:(NSString *)qName
+    attributes:(NSDictionary<NSString *,NSString *> *)attributeDict
+{
+    if ([elementName isEqualToString:@"metadata"]) {
+        _inMetadata = YES;
+    } else if ([elementName isEqualToString:@"manifest"]) {
+        _inManifest = YES;
+    } else if ([elementName isEqualToString:@"spine"]) {
+        _inSpine = YES;
+    }
+
+    if (_inManifest && [elementName isEqualToString:@"item"]) {
+        NSString *itemId   = attributeDict[@"id"];
+        NSString *href     = attributeDict[@"href"];
+        NSString *mediaType = attributeDict[@"media-type"];
+        if (itemId && href) {
+            _manifestItems[itemId] = href;
+            if (mediaType) {
+                _mediaTypes[itemId] = mediaType;
+            }
+        }
+    }
+
+    if (_inSpine && [elementName isEqualToString:@"itemref"]) {
+        NSString *idref = attributeDict[@"idref"];
+        if (idref) {
+            [_spineItemRefs addObject:idref];
+        }
+    }
+
+    _currentText = [NSMutableString string];
+}
+
+- (void)parser:(NSXMLParser *)parser
+foundCharacters:(NSString *)string
+{
+    [_currentText appendString:string];
+}
+
+- (void)parser:(NSXMLParser *)parser
+ didEndElement:(NSString *)elementName
+  namespaceURI:(NSString *)namespaceURI
+ qualifiedName:(NSString *)qName
+{
+    if ([elementName isEqualToString:@"metadata"]) {
+        _inMetadata = NO;
+    } else if ([elementName isEqualToString:@"manifest"]) {
+        _inManifest = NO;
+    } else if ([elementName isEqualToString:@"spine"]) {
+        _inSpine = NO;
+    }
+
+    // Extract <dc:title> from metadata
+    if (_inMetadata && [elementName isEqualToString:@"dc:title"]) {
+        _bookTitle = [_currentText stringByTrimmingCharactersInSet:
+            [NSCharacterSet whitespaceAndNewlineCharacterSet]];
+    }
+
+    _currentText = nil;
+}
+
+@end
+
+#pragma mark - NCX Parser Delegate
+
+/// Parses toc.ncx to extract the table of contents tree.
+@interface WREpubNCXParserDelegate : NSObject <NSXMLParserDelegate>
+@property (nonatomic, strong) NSMutableArray<NSDictionary *> *tocEntries;
+@end
+
+@implementation WREpubNCXParserDelegate
+{
+    BOOL              _inNavPoint;
+    BOOL              _inNavLabel;
+    BOOL              _inContent;
+    BOOL              _inText;
+    NSString         *_currentNavPointId;
+    NSString         *_currentLabel;
+    NSString         *_currentSrc;
+    NSMutableString  *_currentText;
+    NSUInteger        _playOrder;
+}
+
+- (instancetype)init
+{
+    self = [super init];
+    if (self) {
+        _tocEntries = [NSMutableArray array];
+    }
+    return self;
+}
+
+- (void)parser:(NSXMLParser *)parser
+didStartElement:(NSString *)elementName
+  namespaceURI:(NSString *)namespaceURI
+ qualifiedName:(NSString *)qName
+    attributes:(NSDictionary<NSString *,NSString *> *)attributeDict
+{
+    _currentText = [NSMutableString string];
+
+    if ([elementName isEqualToString:@"navPoint"]) {
+        _inNavPoint = YES;
+        _currentNavPointId = attributeDict[@"id"];
+        _playOrder = [attributeDict[@"playOrder"] integerValue];
+    } else if ([elementName isEqualToString:@"navLabel"]) {
+        _inNavLabel = YES;
+    } else if ([elementName isEqualToString:@"text"]) {
+        _inText = YES;
+    } else if ([elementName isEqualToString:@"content"]) {
+        _inContent = YES;
+        _currentSrc = attributeDict[@"src"];
+    }
+}
+
+- (void)parser:(NSXMLParser *)parser
+foundCharacters:(NSString *)string
+{
+    [_currentText appendString:string];
+}
+
+- (void)parser:(NSXMLParser *)parser
+ didEndElement:(NSString *)elementName
+  namespaceURI:(NSString *)namespaceURI
+ qualifiedName:(NSString *)qName
+{
+    if ([elementName isEqualToString:@"text"] && _inText) {
+        _currentLabel = [_currentText stringByTrimmingCharactersInSet:
+            [NSCharacterSet whitespaceAndNewlineCharacterSet]];
+        _inText = NO;
+    } else if ([elementName isEqualToString:@"content"]) {
+        _inContent = NO;
+    } else if ([elementName isEqualToString:@"navLabel"]) {
+        _inNavLabel = NO;
+    } else if ([elementName isEqualToString:@"navPoint"]) {
+        _inNavPoint = NO;
+
+        if (_currentLabel && _currentSrc) {
+            NSDictionary *entry = @{
+                @"id"        : _currentNavPointId ?: @"",
+                @"label"     : _currentLabel,
+                @"src"       : _currentSrc,
+                @"playOrder" : @(_playOrder)
+            };
+            [_tocEntries addObject:entry];
+        }
+
+        _currentNavPointId = nil;
+        _currentLabel      = nil;
+        _currentSrc        = nil;
+    }
+
+    _currentText = nil;
+}
+
+@end