diff options
Diffstat (limited to 'extract/src/odt.c')
| -rw-r--r-- | extract/src/odt.c | 134 |
1 files changed, 83 insertions, 51 deletions
diff --git a/extract/src/odt.c b/extract/src/odt.c index 9e369078..e2e45e2d 100644 --- a/extract/src/odt.c +++ b/extract/src/odt.c @@ -128,7 +128,7 @@ static int s_odt_styles_definitions( ); extract_astring_cat(alloc, out, "<style:paragraph-properties style:writing-mode=\"lr-tb\"/>\n"); extract_astring_cat(alloc, out, "</style:style>\n"); - + /* Style for images. */ extract_astring_cat(alloc, out, "<style:style style:name=\"fr1\" style:family=\"graphic\" style:parent-style-name=\"Graphics\">\n"); extract_astring_cat(alloc, out, "<style:graphic-properties" @@ -156,8 +156,8 @@ static int s_odt_styles_definitions( " draw:color-mode=\"standard\"" "/>\n"); extract_astring_cat(alloc, out, "</style:style>\n"); - - + + if (extract_astring_cat(alloc, out, "</office:automatic-styles>")) return -1; return 0; } @@ -308,9 +308,9 @@ change font. */ if (s_odt_run_finish(alloc, content_state, content)) goto end; } if (s_odt_paragraph_finish(alloc, content)) goto end; - + e = 0; - + end: return e; } @@ -335,14 +335,14 @@ static int s_odt_append_image( ); extract_astring_cat(alloc, content, "</draw:frame>\n"); extract_astring_cat(alloc, content, "</text:p>\n"); - + return 0; } static int s_odt_output_rotated_paragraphs( extract_alloc_t* alloc, - extract_page_t* page, + subpage_t* subpage, int paragraph_begin, int paragraph_end, double rotation_rad, @@ -361,13 +361,13 @@ static int s_odt_output_rotated_paragraphs( int p; double pt_to_inch = 1/72.0; outf("rotated paragraphs: rotation_rad=%f (x y)=(%f %f) (w h)=(%f %f)", rotation_rad, x_pt, y_pt, w_pt, h_pt); - + // https://docs.oasis-open.org/office/OpenDocument/v1.3/cs02/part3-schema/OpenDocument-v1.3-cs02-part3-schema.html#attribute-draw_transform // says rotation is in degrees, but we seem to require -radians. // - + if (!e) e = extract_astring_cat(alloc, content, "\n"); - + if (!e) e = extract_astring_cat(alloc, content, "<text:p text:style-name=\"Standard\">\n"); if (!e) e = extract_astring_catf(alloc, content, "<draw:frame" " text:anchor-type=\"paragraph\"" @@ -388,19 +388,19 @@ static int s_odt_output_rotated_paragraphs( y_pt * pt_to_inch ); if (!e) e = extract_astring_cat(alloc, content, "<draw:text-box>\n"); - + for (p=paragraph_begin; p<paragraph_end; ++p) { - paragraph_t* paragraph = page->paragraphs[p]; + paragraph_t* paragraph = subpage->paragraphs[p]; if (!e) e = s_document_to_odt_content_paragraph(alloc, content_state, paragraph, content, styles); } - + if (!e) e = extract_astring_cat(alloc, content, "\n"); if (!e) e = extract_astring_cat(alloc, content, "</draw:text-box>\n"); if (!e) e = extract_astring_cat(alloc, content, "</draw:frame>\n"); - + if (!e) e = extract_astring_cat(alloc, content, "</text:p>\n"); - + return e; } @@ -409,7 +409,7 @@ static int s_odt_append_table(extract_alloc_t* alloc, table_t* table, extract_as { int e = -1; int y; - + { int x; static int table_number = 0; @@ -438,7 +438,7 @@ static int s_odt_append_table(extract_alloc_t* alloc, table_t* table, extract_as if (extract_astring_cat(alloc, content, " <table:table-row>\n" )) goto end; - + for (x=0; x<table->cells_num_x; ++x) { cell_t* cell = table->cells[y*table->cells_num_x + x]; @@ -447,7 +447,7 @@ static int s_odt_append_table(extract_alloc_t* alloc, table_t* table, extract_as if (extract_astring_cat(alloc, content, " <table:covered-table-cell/>\n")) goto end; continue; } - + if (extract_astring_cat(alloc, content, " <table:table-cell")) goto end; if (cell->extend_right > 1) { @@ -458,7 +458,7 @@ static int s_odt_append_table(extract_alloc_t* alloc, table_t* table, extract_as if (extract_astring_catf(alloc, content, " table:number-rows-spanned=\"%i\"", cell->extend_down)) goto end; } if (extract_astring_catf(alloc, content, ">\n")) goto end; - + /* Write contents of this cell. */ { int p; @@ -482,7 +482,7 @@ static int s_odt_append_table(extract_alloc_t* alloc, table_t* table, extract_as } if (extract_astring_cat(alloc, content, " </table:table>\n")) goto end; e = 0; - + end: return e; } @@ -490,7 +490,7 @@ static int s_odt_append_table(extract_alloc_t* alloc, table_t* table, extract_as static int s_odt_append_rotated_paragraphs( extract_alloc_t* alloc, - extract_page_t* page, + subpage_t* subpage, content_state_t* content_state, int* p, int* text_box_id, @@ -499,7 +499,7 @@ static int s_odt_append_rotated_paragraphs( extract_astring_t* content, extract_odt_styles_t* styles ) -/* Appends paragraphs with same rotation, starting with page->paragraphs[*p] +/* Appends paragraphs with same rotation, starting with subpage->paragraphs[*p] and updates *p. */ { /* Find extent of paragraphs with this same rotation. extent @@ -509,7 +509,7 @@ and updates *p. */ point_t extent = {0, 0}; int p0 = *p; int p1; - paragraph_t* paragraph = page->paragraphs[*p]; + paragraph_t* paragraph = subpage->paragraphs[*p]; outf("rotate=%.2frad=%.1fdeg ctm: ef=(%f %f) abcd=(%f %f %f %f)", rotate, rotate * 180 / pi, @@ -546,9 +546,9 @@ and updates *p. */ ctm->a, ctm->b, ctm->c, ctm->d); } - for (*p=p0; *p<page->paragraphs_num; ++*p) + for (*p=p0; *p<subpage->paragraphs_num; ++*p) { - paragraph = page->paragraphs[*p]; + paragraph = subpage->paragraphs[*p]; ctm = ¶graph->lines[0]->spans[0]->ctm; rotate = atan2(ctm->b, ctm->a); if (rotate != rotate0) @@ -597,7 +597,7 @@ and updates *p. */ if (s_odt_output_rotated_paragraphs( alloc, - page, + subpage, p0, p1, rotate, @@ -612,15 +612,15 @@ and updates *p. */ )) goto end; *p = p1 - 1; e = 0; - + end: return e; } -int extract_document_to_odt_content( +static int extract_page_to_odt_content( extract_alloc_t* alloc, - document_t* document, + extract_page_t* page, int spacing, int rotation, int images, @@ -630,12 +630,12 @@ int extract_document_to_odt_content( { int ret = -1; int text_box_id = 0; - int p; + int c; /* Write paragraphs into <content>. */ - for (p=0; p<document->pages_num; ++p) + for (c=0; c<page->subpages_num; ++c) { - extract_page_t* page = document->pages[p]; + subpage_t* subpage = page->subpages[c]; int p = 0; int t = 0; content_state_t content_state; @@ -644,17 +644,17 @@ int extract_document_to_odt_content( content_state.font.bold = 0; content_state.font.italic = 0; content_state.ctm_prev = NULL; - + for(;;) { - paragraph_t* paragraph = (p == page->paragraphs_num) ? NULL : page->paragraphs[p]; - table_t* table = (t == page->tables_num) ? NULL : page->tables[t]; + paragraph_t* paragraph = (p == subpage->paragraphs_num) ? NULL : subpage->paragraphs[p]; + table_t* table = (t == subpage->tables_num) ? NULL : subpage->tables[t]; double y_paragraph; double y_table; if (!paragraph && !table) break; y_paragraph = (paragraph) ? paragraph->lines[0]->spans[0]->chars[0].y : DBL_MAX; y_table = (table) ? table->pos.y : DBL_MAX; - + if (paragraph && y_paragraph < y_table) { const matrix_t* ctm = ¶graph->lines[0]->spans[0]->ctm; @@ -683,7 +683,7 @@ int extract_document_to_odt_content( if (rotation && rotate != 0) { - if (s_odt_append_rotated_paragraphs(alloc, page, &content_state, &p, &text_box_id, ctm, rotate, content, styles)) goto end; + if (s_odt_append_rotated_paragraphs(alloc, subpage, &content_state, &p, &text_box_id, ctm, rotate, content, styles)) goto end; } else { @@ -697,15 +697,15 @@ int extract_document_to_odt_content( t += 1; } } - + outf("images=%i", images); if (images) { int i; - outf("page->images_num=%i", page->images_num); - for (i=0; i<page->images_num; ++i) + outf("subpage->images_num=%i", subpage->images_num); + for (i=0; i<subpage->images_num; ++i) { - s_odt_append_image(alloc, content, &page->images[i]); + s_odt_append_image(alloc, content, &subpage->images[i]); } } } @@ -716,6 +716,38 @@ int extract_document_to_odt_content( return ret; } +int extract_document_to_odt_content( + extract_alloc_t* alloc, + document_t* document, + int spacing, + int rotation, + int images, + extract_astring_t* content, + extract_odt_styles_t* styles + ) +{ + int p; + int ret = 0; + + /* Write paragraphs into <content>. */ + for (p=0; p<document->pages_num; ++p) + { + extract_page_t* page = document->pages[p]; + + ret = extract_page_to_odt_content( + alloc, + page, + spacing, + rotation, + images, + content, + styles + ); + if (ret) break; + }; + + return ret; +} #if 0 static int s_find_mid(const char* text, const char* begin, const char* end, const char** o_begin, const char** o_end) @@ -749,7 +781,7 @@ int extract_odt_content_item( extract_astring_t temp; extract_astring_init(&temp); *text2 = NULL; - + (void) images; if (0) {} @@ -771,10 +803,10 @@ int extract_odt_content_item( &text_intermediate )) goto end; outf("text_intermediate: %s", text_intermediate); - + /* Convert <styles> to text. */ if (s_odt_styles_definitions(alloc, styles, &styles_definitions)) goto end; - + /* To make tables work, we seem to need to specify table and column styles, and these can be empty. todo: maybe specify exact sizes based on the pdf table and cell dimensions. */ @@ -783,7 +815,7 @@ int extract_odt_content_item( "<style:style style:name=\"extract.table\" style:family=\"table\"/>\n" "<style:style style:name=\"extract.table.column\" style:family=\"table-column\"/>\n" )) goto end; - + /* Replace '<office:automatic-styles/>' with text from <styles_definitions>. */ e = extract_content_insert( @@ -845,7 +877,7 @@ int extract_odt_content_item( return e; } - + int extract_odt_write_template( extract_alloc_t* alloc, @@ -867,7 +899,7 @@ int extract_odt_write_template( assert(path_out); assert(path_template); - + if (extract_check_path_shell_safe(path_out)) { outf("path_out is unsafe: %s", path_out); @@ -896,7 +928,7 @@ int extract_odt_write_template( /* Might be nice to iterate through all items in path_tempdir, but for now we look at just the items that we know extract_odt_content_item() will modify. */ - + { const char* names[] = { @@ -912,7 +944,7 @@ int extract_odt_write_template( extract_free(alloc, &text2); if (extract_asprintf(alloc, &path, "%s/%s", path_tempdir, name) < 0) goto end; if (extract_read_all_path(alloc, path, &text)) goto end; - + outf("before extract_odt_content_item() styles->styles_num=%i", styles->styles_num); if (extract_odt_content_item( alloc, @@ -928,7 +960,7 @@ int extract_odt_write_template( outf("extract_odt_content_item() failed"); goto end; } - + outf("after extract_odt_content_item styles->styles_num=%i", styles->styles_num); { @@ -954,7 +986,7 @@ int extract_odt_write_template( if (extract_asprintf(alloc, &path, "%s/Pictures/%s", path_tempdir, image->name) < 0) goto end; if (extract_write_all(image->data, image->data_size, path)) goto end; } - + outf("Zipping tempdir to create %s", path_out); { const char* path_out_leaf = strrchr(path_out, '/'); |