diff options
Diffstat (limited to 'src/plugins/Fourchan.cpp')
| -rw-r--r-- | src/plugins/Fourchan.cpp | 112 |
1 files changed, 72 insertions, 40 deletions
diff --git a/src/plugins/Fourchan.cpp b/src/plugins/Fourchan.cpp index c91bd0d..7cad54f 100644 --- a/src/plugins/Fourchan.cpp +++ b/src/plugins/Fourchan.cpp @@ -43,7 +43,7 @@ namespace QuickMedia { }; DataView text; // Set when type is TEXT, QUOTE or QUOTELINK - int64_t quote_postnumber; // Set when type is QUOTELINK + int64_t quote_postnumber = 0; // Set when type is QUOTELINK Type type; }; @@ -63,58 +63,90 @@ namespace QuickMedia { return nullptr; } + static void lstrip_newline(const char *str, size_t size, const char **output_str, size_t *output_size) { + size_t i = 0; + while(i < size && str[i] == '\n') { + ++i; + } + *output_str = str + i; + *output_size = size - i; + } + + static void rstrip_newline(const char *str, size_t size, size_t *output_size) { + ssize_t i = size - 1; + while(i >= 0 && str[i] == '\n') { + --i; + } + *output_size = i + 1; + } + + static void strip_newline(const char *str, size_t size, const char **output_str, size_t *output_size) { + lstrip_newline(str, size, output_str, output_size); + rstrip_newline(*output_str, *output_size, output_size); + } + using CommentPieceCallback = std::function<void(const CommentPiece&)>; - static void extract_comment_pieces(TidyDoc doc, TidyNode node, CommentPieceCallback callback) { + static int extract_comment_pieces(TidyDoc doc, TidyNode node, CommentPieceCallback &callback) { for(TidyNode child = tidyGetChild(node); child; child = tidyGetNext(child)) { const char *node_name = tidyNodeGetName(child); - if(node_name && strcmp(node_name, "wbr") == 0) { - CommentPiece comment_piece; - comment_piece.type = CommentPiece::Type::LINE_CONTINUE; - comment_piece.text = { (char*)"", 0 }; - callback(comment_piece); - continue; - } - TidyNodeType node_type = tidyNodeGetType(child); - if(node_type == TidyNode_Start && node_name) { - TidyNode text_node = tidyGetChild(child); - //fprintf(stderr, "Child node name: %s, child text type: %d\n", node_name, tidyNodeGetType(text_node)); - if(tidyNodeGetType(text_node) == TidyNode_Text) { - TidyBuffer tidy_buffer; - tidyBufInit(&tidy_buffer); - if(tidyNodeGetText(doc, text_node, &tidy_buffer)) { + if(node_name) { + if(strcmp(node_name, "br") == 0) { + CommentPiece comment_piece; + comment_piece.type = CommentPiece::Type::TEXT; + // Warning: Cast from const char* to char* ... + comment_piece.text = { (char*)"\n", 1 }; + callback(comment_piece); + /*} else if(strcmp(node_name, "span") == 0) { + const char *span_class = get_attribute_value(child, "class"); + //fprintf(stderr, "span class: %s\n", span_class); + if(span_class && strcmp(span_class, "quote") == 0) { CommentPiece comment_piece; - comment_piece.type = CommentPiece::Type::TEXT; - comment_piece.text = { (char*)tidy_buffer.bp, tidy_buffer.size }; - if(strcmp(node_name, "span") == 0) { - const char *span_class = get_attribute_value(child, "class"); - //fprintf(stderr, "span class: %s\n", span_class); - if(span_class && strcmp(span_class, "quote") == 0) - comment_piece.type = CommentPiece::Type::QUOTE; - } else if(strcmp(node_name, "a") == 0) { - const char *a_class = get_attribute_value(child, "class"); - const char *a_href = get_attribute_value(child, "href"); - //fprintf(stderr, "a class: %s, href: %s\n", a_class, a_href); - if(a_class && a_href && strcmp(a_class, "quotelink") == 0 && strncmp(a_href, "#p", 2) == 0) { - comment_piece.type = CommentPiece::Type::QUOTELINK; - comment_piece.quote_postnumber = strtoll(a_href + 2, nullptr, 10); - } - } + comment_piece.type = CommentPiece::Type::QUOTE; + // Warning: Cast from const char* to char* ... + comment_piece.text = { (char*)"\n", 1 }; callback(comment_piece); - } - tidyBufFree(&tidy_buffer); + }*/ } - } else if(node_type == TidyNode_Text) { + } + + if(tidyNodeGetType(child) == TidyNode_Text) { TidyBuffer tidy_buffer; tidyBufInit(&tidy_buffer); if(tidyNodeGetText(doc, child, &tidy_buffer)) { + const char *inner_text = (const char*)tidy_buffer.bp; + size_t inner_text_size = tidy_buffer.size; + strip_newline(inner_text, inner_text_size, &inner_text, &inner_text_size); + + const char *node_name = tidyNodeGetName(node); + if(node_name && strcmp(node_name, "a") == 0) { + const char *a_class = get_attribute_value(node, "class"); + const char *a_href = get_attribute_value(node, "href"); + if(a_class && a_href && strcmp(a_class, "quotelink") == 0 && strncmp(a_href, "#p", 2) == 0) { + CommentPiece comment_piece; + comment_piece.type = CommentPiece::Type::QUOTELINK; + comment_piece.quote_postnumber = strtoll(a_href + 2, nullptr, 10); + // Warning: Cast from const char* to char* ... + comment_piece.text = { (char*)inner_text, inner_text_size }; + callback(comment_piece); + tidyBufFree(&tidy_buffer); + continue; + } + } + CommentPiece comment_piece; comment_piece.type = CommentPiece::Type::TEXT; - comment_piece.text = { (char*)tidy_buffer.bp, tidy_buffer.size }; + // Warning: Cast from const char* to char* ... + comment_piece.text = { (char*)inner_text, inner_text_size }; callback(comment_piece); } tidyBufFree(&tidy_buffer); + } else { + int res = extract_comment_pieces(doc, child, callback); + if(res != 0) + return res; } } + return 0; } static void extract_comment_pieces(const char *html_source, size_t size, CommentPieceCallback callback) { @@ -129,7 +161,7 @@ namespace QuickMedia { comment_piece.text = { (char*)html_source, size }; callback(comment_piece); } else { - extract_comment_pieces(doc, tidyGetBody(doc), std::move(callback)); + extract_comment_pieces(doc, tidyGetRoot(doc), callback); } tidyRelease(doc); } @@ -241,8 +273,8 @@ namespace QuickMedia { comment_text.append(cp.text.data, cp.text.size); break; case CommentPiece::Type::QUOTE: - comment_text += '>'; - comment_text.append(cp.text.data, cp.text.size); + //comment_text += '>'; + //comment_text.append(cp.text.data, cp.text.size); //comment_text += '\n'; break; case CommentPiece::Type::QUOTELINK: { |