diff options
| -rw-r--r-- | Jamfile | 2 | ||||
| -rw-r--r-- | TODO | 20 | ||||
| -rw-r--r-- | apps/common/pdfapp.c | 23 | ||||
| -rw-r--r-- | fitz/node_tolisp.c (renamed from fitz/node_debug.c) | 0 | ||||
| -rw-r--r-- | fitz/node_toxml.c | 206 | ||||
| -rw-r--r-- | include/mupdf/page.h | 4 | ||||
| -rw-r--r-- | mupdf/pdf_unicode.c | 48 |
7 files changed, 273 insertions, 30 deletions
@@ -98,7 +98,7 @@ Library libfitz : base_rect.c base_rune.c - node_debug.c + node_toxml.c node_misc1.c node_misc2.c node_optimize.c @@ -1,10 +1,24 @@ cmap one-to-many mapping +builtin standard cmap files +put unicode strings in text object, not font + +xml parser +unicode normaliser + +path stroke/dash/flatten work on real path struct +turn into gel as second step after stroke/flatten +add intersector for metro union/xor/difference stuff + +image rescale to exact size instead of by integer quantas + +public / private api + +top-level "driver" architecture (metro/pdf/whatever input) + +--- immediate plan: - * design gui for editor - * libfontfocus - * write mupdf hackers guide * clean up and 'freeze' public api * get font bbox from fontdescriptor if available diff --git a/apps/common/pdfapp.c b/apps/common/pdfapp.c index d7bed118..ef342b43 100644 --- a/apps/common/pdfapp.c +++ b/apps/common/pdfapp.c @@ -394,6 +394,15 @@ void pdfapp_onkey(pdfapp_t *app, int c) pdfapp_showpage(app, 0, 1); break; + case 'a': + app->rotate -= 15; + pdfapp_showpage(app, 0, 1); + break; + case 's': + app->rotate += 15; + pdfapp_showpage(app, 0, 1); + break; + /* * Pan view, but dont need to repaint image */ @@ -602,9 +611,11 @@ void pdfapp_oncopy(pdfapp_t *app, unsigned short *ucsbuf, int ucslen) { fz_error *error; pdf_textline *line, *ln; - int x, y, c; + int y, c; int i, p; + int bx0, bx1, by0, by1; + int x0 = app->image->x + app->selr.x0 - app->panx; int x1 = app->image->x + app->selr.x1 - app->panx; int y0 = app->image->y + app->selr.y0 - app->pany; @@ -620,17 +631,19 @@ void pdfapp_oncopy(pdfapp_t *app, unsigned short *ucsbuf, int ucslen) y = y0 - 1; for (i = 0; i < ln->len; i++) { - x = ln->text[i].x; - y = ln->text[i].y; + bx0 = ln->text[i].bbox.x0; + bx1 = ln->text[i].bbox.x1; + by0 = ln->text[i].bbox.y0; + by1 = ln->text[i].bbox.y1; c = ln->text[i].c; if (c < 32) c = '?'; - if (x >= x0 && x <= x1 && y >= y0 && y <= y1) + if (bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1) if (p < ucslen - 1) ucsbuf[p++] = c; } - if (y >= y0 && y <= y1) + if (by1 >= y0 && by0 <= y1) { #ifdef WIN32 if (p < ucslen - 1) diff --git a/fitz/node_debug.c b/fitz/node_tolisp.c index b03d4e28..b03d4e28 100644 --- a/fitz/node_debug.c +++ b/fitz/node_tolisp.c diff --git a/fitz/node_toxml.c b/fitz/node_toxml.c new file mode 100644 index 00000000..04a00423 --- /dev/null +++ b/fitz/node_toxml.c @@ -0,0 +1,206 @@ +#include <fitz.h> + +static void indent(int level) +{ + while (level--) + putchar(' '); +} + +static void xmlnode(fz_node *node, int level); + +static void xmlmeta(fz_metanode *node, int level) +{ + fz_node *child; + + indent(level); + printf("<meta"); + if (node->name) + { + printf(" name=\""); + fz_debugobj(node->name); + printf("\" "); + } + printf(">\n"); + + for (child = node->super.first; child; child = child->next) + xmlnode(child, level + 1); + + indent(level); + printf("</meta>\n"); +} + +static void xmlover(fz_overnode *node, int level) +{ + fz_node *child; + indent(level); + printf("<over>\n"); + for (child = node->super.first; child; child = child->next) + xmlnode(child, level + 1); + indent(level); + printf("</over>\n"); +} + +static void xmlmask(fz_masknode *node, int level) +{ + fz_node *child; + indent(level); + printf("<mask>\n"); + for (child = node->super.first; child; child = child->next) + xmlnode(child, level + 1); + indent(level); + printf("</mask>\n"); +} + +static void xmlblend(fz_blendnode *node, int level) +{ + fz_node *child; + indent(level); + printf("<blend mode=\"%d\">\n", node->mode); + for (child = node->super.first; child; child = child->next) + xmlnode(child, level + 1); + indent(level); + printf("</blend>\n"); +} + +static void xmltransform(fz_transformnode *node, int level) +{ + indent(level); + printf("<transform matrix=\"%g %g %g %g %g %g\">\n", + node->m.a, node->m.b, + node->m.c, node->m.d, + node->m.e, node->m.f); + xmlnode(node->super.first, level + 1); + indent(level); + printf("</transform>\n"); +} + +static void xmlcolor(fz_colornode *node, int level) +{ + int i; + indent(level); + printf("<solid colorspace=\"%s\" v=\"", node->cs->name); + for (i = 0; i < node->n; i++) + { + printf("%g", node->samples[i]); + if (i < node->n - 1) + putchar(' '); + } + printf("\" />\n"); +} + +static void xmllink(fz_linknode *node, int level) +{ + indent(level); + printf("<link name=\"%p\" />\n", node->tree); +} + +static void xmlpath(fz_pathnode *node, int level) +{ + int i; + + indent(level); + + if (node->paint == FZ_STROKE) + { + printf("<path fill=\"stroke\" cap=\"%d\" join=\"%d\" width=\"%g\" miter=\"%g\"", + node->linecap, + node->linejoin, + node->linewidth, + node->miterlimit); + if (node->dash) + { + printf(" phase=\"%g\" array=\"", node->dash->phase); + for (i = 0; i < node->dash->len; i++) + printf("%g ", node->dash->array[i]); + printf("\""); + } + printf(">\n"); + } + else + { + printf("<path fill=\"%s\">\n", + node->paint == FZ_FILL ? "nonzero" : "evenodd"); + } + + fz_debugpathnode(node); + + indent(level); + printf("</path>\n"); +} + +static void xmltext(fz_textnode *node, int level) +{ + int i; + + indent(level); + printf("<text font=\"%s\" matrix=\"%g %g %g %g\">\n", node->font->name, + node->trm.a, node->trm.b, node->trm.c, node->trm.d); + + for (i = 0; i < node->len; i++) + { + indent(level + 1); + if (node->els[i].cid >= 32 && node->els[i].cid < 128) + printf("<g c=\"%c\" x=\"%g\" y=\"%g\" />\n", + node->els[i].cid, node->els[i].x, node->els[i].y); + else + printf("<g c=\"<%04x>\" x=\"%g\" y=\"%g\" />\n", + node->els[i].cid, node->els[i].x, node->els[i].y); + } + + indent(level); + printf("</text>\n"); +} + +static void xmlimage(fz_imagenode *node, int level) +{ + fz_image *image = node->image; + indent(level); + printf("<image w=\"%d\" h=\"%d\" n=\"%d\" a=\"%d\" />\n", + image->w, image->h, image->n, image->a); +} + +static void xmlshade(fz_shadenode *node, int level) +{ + indent(level); + printf("<shade />\n"); +} + +static void xmlnode(fz_node *node, int level) +{ + if (!node) + { + indent(level); + printf("<nil />\n"); + return; + } + + switch (node->kind) + { + case FZ_NMETA: xmlmeta((fz_metanode*)node, level); break; + case FZ_NOVER: xmlover((fz_overnode*)node, level); break; + case FZ_NMASK: xmlmask((fz_masknode*)node, level); break; + case FZ_NBLEND: xmlblend((fz_blendnode*)node, level); break; + case FZ_NTRANSFORM: xmltransform((fz_transformnode*)node, level); break; + case FZ_NCOLOR: xmlcolor((fz_colornode*)node, level); break; + case FZ_NPATH: xmlpath((fz_pathnode*)node, level); break; + case FZ_NTEXT: xmltext((fz_textnode*)node, level); break; + case FZ_NIMAGE: xmlimage((fz_imagenode*)node, level); break; + case FZ_NSHADE: xmlshade((fz_shadenode*)node, level); break; + case FZ_NLINK: xmllink((fz_linknode*)node, level); break; + } +} + +void +fz_debugnode(fz_node *node) +{ + xmlnode(node, 0); +} + +void +fz_debugtree(fz_tree *tree) +{ + printf("<tree>\n"); + xmlnode(tree->root, 1); + printf("</tree>\n"); +} + diff --git a/include/mupdf/page.h b/include/mupdf/page.h index 4304fa22..57aae34a 100644 --- a/include/mupdf/page.h +++ b/include/mupdf/page.h @@ -27,12 +27,12 @@ struct pdf_page_s struct pdf_textchar_s { - int x, y, c; + fz_irect bbox; + int c; }; struct pdf_textline_s { - fz_point height; int len, cap; pdf_textchar *text; pdf_textline *next; diff --git a/mupdf/pdf_unicode.c b/mupdf/pdf_unicode.c index c9cc0807..1df96667 100644 --- a/mupdf/pdf_unicode.c +++ b/mupdf/pdf_unicode.c @@ -99,8 +99,6 @@ pdf_newtextline(pdf_textline **linep) line = *linep = fz_malloc(sizeof(pdf_textline)); if (!line) return fz_outofmem; - line->height.x = 0; /* bad default value... */ - line->height.y = 10; line->len = 0; line->cap = 0; line->text = nil; @@ -118,7 +116,7 @@ pdf_droptextline(pdf_textline *line) } static fz_error * -addtextchar(pdf_textline *line, int x, int y, int c) +addtextchar(pdf_textline *line, fz_irect bbox, int c) { pdf_textchar *newtext; int newcap; @@ -133,8 +131,7 @@ addtextchar(pdf_textline *line, int x, int y, int c) line->text = newtext; } - line->text[line->len].x = x; - line->text[line->len].y = y; + line->text[line->len].bbox = bbox; line->text[line->len].c = c; line->len ++; @@ -158,19 +155,15 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm) fz_matrix trm; float dx, dy, t; fz_point p; + fz_point vx; + fz_point vy; fz_vmtx v; fz_hmtx h; - int i, g, x, y; + int i, g; + int x, y; + fz_irect box; int c; - /* get line height */ - trm = fz_concat(tm, ctm); - trm.e = 0; - trm.f = 0; - p.x = 0; - p.y = 1; - (*line)->height = fz_transformpoint(trm, p); - for (i = 0; i < text->len; i++) { g = text->els[i].cid; @@ -178,8 +171,10 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm) tm.e = text->els[i].x; tm.f = text->els[i].y; trm = fz_concat(tm, ctm); - x = fz_floor(trm.e); - y = fz_floor(trm.f); + x = trm.e; + y = trm.f; + trm.e = 0; + trm.f = 0; p.x = text->els[i].x; p.y = text->els[i].y; @@ -192,12 +187,18 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm) { h = fz_gethmtx(text->font, g); oldpt.x += h.w * 0.001; + + vx.x = h.w * 0.001; vx.y = 0; + vy.x = 0; vy.y = 1; } else { v = fz_getvmtx(text->font, g); - oldpt.y += v.w; + oldpt.y += v.w * 0.001; t = dy; dy = dx; dx = t; + + vx.x = 0.5; vx.y = 0; + vy.x = 0; vy.y = v.w * 0.001; } if (fabs(dy) > 0.2) @@ -211,11 +212,20 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm) } else if (fabs(dx) > 0.2) { - error = addtextchar(*line, x, y, ' '); + box.x0 = x; box.x1 = x; + box.y0 = y; box.y1 = y; + error = addtextchar(*line, box, ' '); if (error) return error; } + vx = fz_transformpoint(trm, vx); + vy = fz_transformpoint(trm, vy); + box.x0 = MIN(0, MIN(vx.x, vy.x)) + x; + box.x1 = MAX(0, MAX(vx.x, vy.x)) + x; + box.y0 = MIN(0, MIN(vx.y, vy.y)) + y; + box.y1 = MAX(0, MAX(vx.y, vy.y)) + y; + if (font->tounicode) c = fz_lookupcid(font->tounicode, g); else if (g < font->ncidtoucs) @@ -223,7 +233,7 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm) else c = g; - error = addtextchar(*line, x, y, c); + error = addtextchar(*line, box, c); if (error) return error; } |