--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/mupdf-source/thirdparty/extract/src/odt.c	Mon Sep 15 11:43:07 2025 +0200
@@ -0,0 +1,1069 @@
+/* These extract_odt_*() functions generate odt content and odt zip archive
+data.
+
+Caller must call things in a sensible order to create valid content -
+e.g. don't call odt_paragraph_start() twice without intervening call to
+odt_paragraph_finish(). */
+
+#include "extract/extract.h"
+
+#include "odt_template.h"
+
+#include "astring.h"
+#include "document.h"
+#include "odt.h"
+#include "mem.h"
+#include "memento.h"
+#include "outf.h"
+#include "sys.h"
+#include "text.h"
+#include "zip.h"
+
+#include <assert.h>
+#include <errno.h>
+#include <float.h>
+#include <math.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+#include <sys/stat.h>
+
+
+static int
+odt_paragraph_start(
+		extract_alloc_t   *alloc,
+		extract_astring_t *output)
+{
+	return extract_astring_cat(alloc, output, "\n\n<text:p>");
+}
+
+static int
+odt_paragraph_finish(
+		extract_alloc_t   *alloc,
+		extract_astring_t *output)
+{
+	return extract_astring_cat(alloc, output, "</text:p>");
+}
+
+/* ODT doesn't seem to support ad-hoc inline font specifications; instead
+we have to define a style at the start of the content.xml file. So when
+writing content we insert a style name and add the required styles to a
+extract_odt_styles_t struct. */
+
+struct extract_odt_style_t
+{
+	int     id; /* A unique id for this style. */
+	font_t  font;
+};
+
+struct extract_odt_styles_t
+{
+	/* Styles are stored sorted. */
+	extract_odt_style_t *styles;
+	int                  styles_num;
+};
+
+static int
+odt_style_compare(
+		extract_odt_style_t *a,
+		extract_odt_style_t *b)
+{
+	int d;
+	double dd;
+
+	if ((d = strcmp(a->font.name, b->font.name)))   return d;
+	if ((dd = a->font.size - b->font.size) != 0.0)  return (dd > 0.0) ? 1 : -1;
+	if ((d = a->font.bold - b->font.bold))          return d;
+	if ((d = a->font.italic - b->font.italic))      return d;
+
+	return 0;
+}
+
+static int
+odt_style_append_definition(
+		extract_alloc_t     *alloc,
+		extract_odt_style_t *style,
+		extract_astring_t   *text)
+{
+	const char* font_name = style->font.name;
+
+	/* This improves output e.g. for zlib.3.pdf, but clearly a hack. */
+	if (0 && strstr(font_name, "Helvetica"))
+	{
+		font_name = "Liberation Sans";
+	}
+	outf("style->font_name=%s font_name=%s", style->font.name, font_name);
+	if (extract_astring_catf(alloc, text, "<style:style style:name=\"T%i\" style:family=\"text\">", style->id)) return -1;
+	if (extract_astring_catf(alloc, text, "<style:text-properties style:font-name=\"%s\"", font_name)) return -1;
+	if (extract_astring_catf(alloc, text, " fo:font-size=\"%.2fpt\"", style->font.size)) return -1;
+	if (extract_astring_catf(alloc, text, " fo:font-weight=\"%s\"", style->font.bold ? "bold" : "normal")) return -1;
+	if (extract_astring_catf(alloc, text, " fo:font-style=\"%s\"", style->font.italic ? "italic" : "normal")) return -1;
+	if (extract_astring_cat(alloc, text, " /></style:style>")) return -1;
+
+	return 0;
+}
+
+void
+extract_odt_styles_free(
+		extract_alloc_t      *alloc,
+		extract_odt_styles_t *styles)
+{
+	int i;
+
+	for (i=0; i<styles->styles_num; ++i)
+	{
+		extract_odt_style_t* style = &styles->styles[i];
+		extract_free(alloc, &style->font.name);
+	}
+	extract_free(alloc, &styles->styles);
+}
+
+static int
+odt_styles_definitions(
+		extract_alloc_t      *alloc,
+		extract_odt_styles_t *styles,
+		extract_astring_t    *out)
+{
+	int i;
+
+	if (extract_astring_cat(alloc, out, "<office:automatic-styles>")) return -1;
+	for (i=0; i<styles->styles_num; ++i)
+	{
+		if (odt_style_append_definition(alloc, &styles->styles[i], out)) return -1;
+	}
+	extract_astring_cat(alloc, out, "<style:style style:name=\"gr1\" style:family=\"graphic\">\n");
+	extract_astring_cat(alloc, out, "<style:graphic-properties"
+			" draw:stroke=\"none\""
+			" svg:stroke-color=\"#000000\""
+			" draw:fill=\"none\""
+			" draw:fill-color=\"#ffffff\""
+			" fo:min-height=\"1.9898in\""
+			" style:run-through=\"foreground\""
+			" style:wrap=\"run-through\""
+			" style:number-wrapped-paragraphs=\"no-limit\""
+			" style:vertical-pos=\"from-top\""
+			" style:vertical-rel=\"paragraph\""
+			" style:horizontal-pos=\"from-left\""
+			" style:horizontal-rel=\"paragraph\""
+			" />\n"
+			);
+	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"
+			" fo:margin-left=\"0in\""
+			" fo:margin-right=\"0in\""
+			" fo:margin-top=\"0in\""
+			" fo:margin-bottom=\"0in\""
+			" style:vertical-pos=\"top\""
+			" style:vertical-rel=\"baseline\""
+			" fo:background-color=\"transparent\""
+			" draw:fill=\"none\""
+			" draw:fill-color=\"#ffffff\""
+			" fo:padding=\"0in\""
+			" fo:border=\"none\""
+			" style:mirror=\"none\""
+			" fo:clip=\"rect(0in, 0in, 0in, 0in)\""
+			" draw:luminance=\"0%\""
+			" draw:contrast=\"0%\""
+			" draw:red=\"0%\""
+			" draw:green=\"0%\""
+			" draw:blue=\"0%\""
+			" draw:gamma=\"100%\""
+			" draw:color-inversion=\"false\""
+			" draw:image-opacity=\"100%\""
+			" 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;
+}
+
+/* Adds specified style to <styles> if not already present. Sets *o_style to
+point to the style_t within <styles>. */
+static int
+odt_styles_add(
+	extract_alloc_t      *alloc,
+	extract_odt_styles_t *styles,
+	font_t               *font,
+	extract_odt_style_t **o_style)
+{
+	extract_odt_style_t style = {0 /*id*/, *font};
+	int i;
+
+	/* We keep styles->styles[] sorted; todo: use bsearch or similar when
+	searching. */
+	for (i=0; i<styles->styles_num; ++i)
+	{
+		int d = odt_style_compare(&style, &styles->styles[i]);
+		if (d == 0)
+		{
+			*o_style = &styles->styles[i];
+			return 0;
+		}
+		if (d > 0) break;
+	}
+	/* Insert at position <i>. */
+	if (extract_realloc(alloc, &styles->styles, sizeof(styles->styles[0]) * (styles->styles_num+1))) return -1;
+	memmove(&styles->styles[i+1], &styles->styles[i], sizeof(styles->styles[0]) * (styles->styles_num - i));
+	styles->styles_num += 1;
+	styles->styles[i].id = styles->styles_num + 10; /* Leave space for template's built-in styles. */
+	if (extract_strdup(alloc, font->name, &styles->styles[i].font.name)) return -1;
+	styles->styles[i].font.size = font->size;
+	styles->styles[i].font.bold = font->bold;
+	styles->styles[i].font.italic = font->italic;
+	*o_style = &styles->styles[i];
+
+	return 0;
+}
+
+/* Starts a new run. Caller must ensure that s_odt_run_finish() was
+called to terminate any previous run. */
+static int
+extract_odt_run_start(
+		extract_alloc_t      *alloc,
+		extract_astring_t    *content,
+		extract_odt_styles_t *styles,
+		content_state_t      *content_state)
+{
+	extract_odt_style_t* style;
+
+	if (odt_styles_add(alloc,
+					   styles,
+					   &content_state->font,
+					   &style)) return -1;
+	if (extract_astring_catf(alloc, content, "<text:span text:style-name=\"T%i\">", style->id)) return -1;
+
+	return 0;
+}
+
+static int
+odt_run_finish(
+	extract_alloc_t   *alloc,
+	content_state_t   *content_state,
+	extract_astring_t *content)
+{
+	if (content_state)
+		content_state->font.name = NULL;
+	return extract_astring_cat(alloc, content, "</text:span>");
+}
+
+/* Append an empty paragraph to *content. */
+static int
+odt_append_empty_paragraph(
+		extract_alloc_t      *alloc,
+		extract_astring_t    *content,
+		extract_odt_styles_t *styles)
+{
+	int e = -1;
+	static char fontname[] = "OpenSans";
+	content_state_t content_state = {0};
+
+	if (odt_paragraph_start(alloc, content)) goto end;
+	/* [This comment is from docx, haven't checked odt.] It seems like our
+	choice of font size here doesn't make any difference to the amount of
+	vertical space, unless we include a non-space character. Presumably
+	something to do with the styles in the template document. */
+	content_state.font.name = fontname;
+	content_state.font.size = 10;
+	content_state.font.bold = 0;
+	content_state.font.italic = 0;
+	if (extract_odt_run_start(alloc, content, styles, &content_state)) goto end;
+	//docx_char_append_string(content, "&#160;");   /* &#160; is non-break space. */
+	if (odt_run_finish(alloc, NULL /*content_state*/, content)) goto end;
+	if (odt_paragraph_finish(alloc, content)) goto end;
+	e = 0;
+
+end:
+	return e;
+}
+
+
+/* Append odt xml for <paragraph> to <content>. Updates *content_state if we
+change font. */
+static int
+document_to_odt_content_paragraph(
+		extract_alloc_t      *alloc,
+		content_state_t      *content_state,
+		paragraph_t          *paragraph,
+		extract_astring_t    *content,
+		extract_odt_styles_t *styles)
+{
+	int e = -1;
+	content_line_iterator  lit;
+	line_t                *line;
+
+	if (odt_paragraph_start(alloc, content)) goto end;
+
+	/* Output justification. */
+	if ((paragraph->line_flags & paragraph_not_fully_justified) == 0)
+	{
+		if (extract_astring_cat(alloc, content, "<w:pPr><w:jc w:val=\"both\"/></w:pPr>")) goto end;
+	}
+	else if ((paragraph->line_flags & paragraph_not_centred) == 0)
+	{
+		if (extract_astring_cat(alloc, content, "<w:pPr><w:jc w:val=\"center\"/></w:pPr>")) goto end;
+	}
+	else if ((paragraph->line_flags & (paragraph_not_aligned_left | paragraph_not_aligned_right)) == paragraph_not_aligned_left)
+	{
+		if (extract_astring_cat(alloc, content, "<w:pPr><w:jc w:val=\"right\"/></w:pPr>")) goto end;
+	}
+	else if ((paragraph->line_flags & (paragraph_not_aligned_left | paragraph_not_aligned_right)) == paragraph_not_aligned_right)
+	{
+		if (extract_astring_cat(alloc, content, "<w:pPr><w:jc w:val=\"left\"/></w:pPr>")) goto end;
+	}
+
+
+	for (line = content_line_iterator_init(&lit, &paragraph->content); line != NULL; line = content_line_iterator_next(&lit))
+	{
+		content_span_iterator  sit;
+		span_t                *span;
+
+		for (span = content_span_iterator_init(&sit, &line->content); span != NULL; span = content_span_iterator_next(&sit))
+		{
+			int si;
+			double font_size_new;
+
+			content_state->ctm_prev = &span->ctm;
+			font_size_new = extract_font_size(&span->ctm);
+			if (!content_state->font.name
+					|| strcmp(span->font_name, content_state->font.name)
+					|| span->flags.font_bold != content_state->font.bold
+					|| span->flags.font_italic != content_state->font.italic
+					|| font_size_new != content_state->font.size
+					)
+			{
+				if (content_state->font.name)
+				{
+					if (odt_run_finish(alloc, content_state, content)) goto end;
+				}
+				content_state->font.name = span->font_name;
+				content_state->font.bold = span->flags.font_bold;
+				content_state->font.italic = span->flags.font_italic;
+				content_state->font.size = font_size_new;
+				if (extract_odt_run_start( alloc, content, styles, content_state)) goto end;
+			}
+
+			for (si=0; si<span->chars_num; ++si)
+			{
+				char_t* char_ = &span->chars[si];
+				int c = char_->ucs;
+				if (extract_astring_catc_unicode_xml(alloc, content, c)) goto end;
+			}
+			/* Remove any trailing '-' at end of line. */
+			if (extract_astring_char_truncate_if(content, '-')) goto end;
+		}
+		if (paragraph->line_flags & paragraph_breaks_strangely)
+		{
+			if (extract_astring_cat(alloc, content, "<w:br/>")) goto end;
+		}
+	}
+	if (content_state->font.name)
+	{
+		if (odt_run_finish(alloc, content_state, content)) goto end;
+	}
+	if (odt_paragraph_finish(alloc, content)) goto end;
+
+	e = 0;
+
+	end:
+	return e;
+}
+
+/* Write reference to image into odt content. */
+static int
+odt_append_image(
+		extract_alloc_t   *alloc,
+		extract_astring_t *output,
+		image_t           *image)
+{
+	extract_astring_cat(alloc, output, "\n");
+	extract_astring_cat(alloc, output, "<text:p text:style-name=\"Standard\">\n");
+	extract_astring_catf(alloc, output, "<draw:frame draw:style-name=\"fr1\" draw:name=\"Picture %s\" text:anchor-type=\"as-char\" svg:width=\"%fin\" svg:height=\"%fin\" draw:z-index=\"0\">\n",
+				image->id,
+				image->w / 72.0,
+				image->h / 72.0);
+	extract_astring_catf(alloc, output, "<draw:image xlink:href=\"Pictures/%s\" xlink:type=\"simple\" xlink:show=\"embed\" xlink:actuate=\"onLoad\" draw:mime-type=\"image/%s\"/>\n",
+			image->name,
+			image->type);
+	extract_astring_cat(alloc, output, "</draw:frame>\n");
+	extract_astring_cat(alloc, output, "</text:p>\n");
+
+	return 0;
+}
+
+
+/* Writes paragraph to content inside rotated text box. */
+static int
+odt_output_rotated_paragraphs(
+		extract_alloc_t      *alloc,
+		block_t              *block,
+		double                rotation_rad,
+		double                x_pt,
+		double                y_pt,
+		double                w_pt,
+		double                h_pt,
+		int                   text_box_id,
+		extract_astring_t    *content,
+		extract_odt_styles_t *styles,
+		content_state_t      *content_state)
+{
+	int                          e = 0;
+	paragraph_t                *paragraph;
+	content_paragraph_iterator  pit;
+	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\""
+			" draw:z-index=\"5\""
+			" draw:name=\"Shape%i\""
+			" draw:style-name=\"gr1\""
+			" draw:text-style-name=\"Standard\""
+			" svg:width=\"%fin\""
+			" svg:height=\"%fin\""
+			" draw:transform=\"rotate (%f) translate (%fin %fin)\""
+			">\n"
+			,
+			text_box_id,
+			w_pt * pt_to_inch,
+			h_pt * pt_to_inch,
+			-rotation_rad,
+			x_pt * pt_to_inch,
+			y_pt * pt_to_inch
+			);
+	if (!e) e = extract_astring_cat(alloc, content, "<draw:text-box>\n");
+
+	for (paragraph = content_paragraph_iterator_init(&pit, &block->content); paragraph != NULL; paragraph = content_paragraph_iterator_next(&pit))
+		if (!e) e = 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;
+}
+
+
+static int
+odt_append_table(
+		extract_alloc_t      *alloc,
+		table_t              *table,
+		extract_astring_t    *output,
+		extract_odt_styles_t *styles)
+{
+	int e = -1;
+	int y;
+
+	{
+		int x;
+		static int table_number = 0;
+		table_number += 1;
+		if (extract_astring_catf(alloc, output,
+				"\n"
+				"    <table:table text:style-name=\"extract.table\" table:name=\"extract.table.%i\">\n"
+				"        <table:table-columns>\n"
+				,
+				table_number
+				)) goto end;
+
+		for (x=0; x<table->cells_num_x; ++x)
+		{
+			if (extract_astring_cat(alloc, output,
+					"            <table:table-column table:style-name=\"extract.table.column\"/>\n"
+					)) goto end;
+		}
+		if (extract_astring_cat(alloc, output,
+				"        </table:table-columns>\n"
+				)) goto end;
+	}
+	for (y=0; y<table->cells_num_y; ++y)
+	{
+		int x;
+		if (extract_astring_cat(alloc, output,
+				"        <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];
+			content_paragraph_iterator  pit;
+			paragraph_t                *paragraph;
+			content_state_t             content_state;
+
+			if (!cell->above || !cell->left)
+			{
+				if (extract_astring_cat(alloc, output, "            <table:covered-table-cell/>\n")) goto end;
+				continue;
+			}
+
+			if (extract_astring_cat(alloc, output, "            <table:table-cell")) goto end;
+			if (cell->extend_right > 1)
+			{
+				if (extract_astring_catf(alloc, output, " table:number-columns-spanned=\"%i\"", cell->extend_right)) goto end;
+			}
+			if (cell->extend_down > 1)
+			{
+				if (extract_astring_catf(alloc, output, " table:number-rows-spanned=\"%i\"", cell->extend_down)) goto end;
+			}
+			if (extract_astring_catf(alloc, output, ">\n")) goto end;
+
+			/* Write contents of this cell. */
+			content_state.font.name = NULL;
+			content_state.ctm_prev = NULL;
+			for (paragraph = content_paragraph_iterator_init(&pit, &cell->content); paragraph != NULL; paragraph = content_paragraph_iterator_next(&pit))
+				if (document_to_odt_content_paragraph(alloc, &content_state, paragraph, output, styles)) goto end;
+			if (content_state.font.name)
+				if (odt_run_finish(alloc, &content_state, output)) goto end;
+			if (extract_astring_cat(alloc, output, "\n")) goto end;
+			if (extract_astring_cat(alloc, output, "            </table:table-cell>\n")) goto end;
+		}
+		if (extract_astring_cat(alloc, output, "        </table:table-row>\n")) goto end;
+	}
+	if (extract_astring_cat(alloc, output, "    </table:table>\n")) goto end;
+	e = 0;
+
+	end:
+	return e;
+}
+
+
+/* Appends paragraphs with same rotation, starting with subpage->paragraphs[*p]
+and updates *p. */
+static int
+odt_append_rotated_paragraphs(
+		extract_alloc_t       *alloc,
+		content_state_t       *content_state,
+		block_t               *block,
+		int                   *text_box_id,
+		const matrix4_t       *ctm,
+		double                 rotate,
+		extract_astring_t     *output,
+		extract_odt_styles_t  *styles)
+{
+	/* Find extent of paragraphs with this same rotation. extent
+	will contain max width and max height of paragraphs, in units
+	before application of ctm, i.e. before rotation. */
+	int               e           = -1;
+	point_t           extent      = {0, 0};
+	content_iterator  cit;
+	content_t        *content;
+	paragraph_t      *paragraph   = content_first_paragraph(&block->content);
+
+	/* We assume that first span is at origin of text
+	 * block. This assumes left-to-right text. */
+	span_t           *first_span  = content_first_span(&content_first_line(&paragraph->content)->content);
+	point_t           origin      = { first_span->chars[0].x,
+									 first_span->chars[0].y };
+	matrix_t          ctm_inverse = {1, 0, 0, 1, 0, 0};
+	double            ctm_det     = ctm->a*ctm->d - ctm->b*ctm->c;
+
+	outf("rotate=%.2frad=%.1fdeg ctm: origin=(%f %f) abcd=(%f %f %f %f)",
+		 rotate, rotate * 180 / pi,
+		 origin.x,
+		 origin.y,
+		 ctm->a,
+		 ctm->b,
+		 ctm->c,
+		 ctm->d
+		 );
+
+	if (ctm_det != 0)
+	{
+		ctm_inverse.a = +ctm->d / ctm_det;
+		ctm_inverse.b = -ctm->b / ctm_det;
+		ctm_inverse.c = -ctm->c / ctm_det;
+		ctm_inverse.d = +ctm->a / ctm_det;
+	}
+	else
+	{
+		outf("cannot invert ctm=(%f %f %f %f)",
+			 ctm->a, ctm->b, ctm->c, ctm->d);
+	}
+
+	for (content = content_iterator_init(&cit, &block->content); content != NULL; content = content_iterator_next(&cit))
+	{
+		content_line_iterator  lit;
+		line_t                *line;
+		paragraph_t           *paragraph;
+
+		assert(content->type == content_paragraph);
+		if (content->type != content_paragraph)
+			continue; /* This shouldn't happen for now! */
+
+		paragraph = (paragraph_t *)content;
+
+		/* Update <extent>. */
+		for (line = content_line_iterator_init(&lit, &paragraph->content); line != NULL; line = content_line_iterator_next(&lit))
+		{
+			span_t *span = extract_line_span_last(line);
+			char_t *char_ = extract_span_char_last(span);
+			double  adv = char_->adv * extract_font_size(&span->ctm);
+			double  x = char_->x + adv * cos(rotate);
+			double  y = char_->y + adv * sin(rotate);
+
+			double  dx = x - origin.x;
+			double  dy = y - origin.y;
+
+			/* Position relative to origin and before box rotation. */
+			double  xx = ctm_inverse.a * dx + ctm_inverse.b * dy;
+			double  yy = ctm_inverse.c * dx + ctm_inverse.d * dy;
+			yy = -yy;
+			if (xx > extent.x) extent.x = xx;
+			if (yy > extent.y) extent.y = yy;
+			if (0) outf("rotate=%f origin=(%f %f) xy=(%f %f) dxy=(%f %f) xxyy=(%f %f) span: %s",
+						rotate, origin.x, origin.y, x, y, dx, dy, xx, yy, extract_span_string(alloc, span));
+		}
+	}
+	outf("rotate=%f extent is: (%f %f)",
+		 rotate, extent.x, extent.y);
+
+	/* All the paragraphs have same rotation. We output them into
+	 * a single rotated text box. */
+
+	/* We need unique id for text box. */
+	*text_box_id += 1;
+
+	if (odt_output_rotated_paragraphs(
+			alloc,
+			block,
+			rotate,
+			origin.x,
+			origin.y,
+			extent.x,
+			extent.y,
+			*text_box_id,
+			output,
+			styles,
+			content_state)) goto end;
+
+	e = 0;
+end:
+
+	return e;
+}
+
+
+static int
+extract_page_to_odt_content(
+		extract_alloc_t      *alloc,
+		extract_page_t       *page,
+		int                   spacing,
+		int                   rotation,
+		int                   images,
+		extract_astring_t    *output,
+		extract_odt_styles_t *styles)
+{
+	int ret = -1;
+	int text_box_id = 0;
+	int c;
+
+	/* Write paragraphs into <content>. */
+	for (c=0; c<page->subpages_num; ++c)
+	{
+		subpage_t                  *subpage = page->subpages[c];
+		content_iterator            cit;
+		content_t                  *content;
+		content_table_iterator      tit;
+		table_t                    *table;
+		content_state_t content_state;
+		content_state.font.name = NULL;
+		content_state.font.size = 0;
+		content_state.font.bold = 0;
+		content_state.font.italic = 0;
+		content_state.ctm_prev = NULL;
+
+		content = content_iterator_init(&cit, &subpage->content);
+		table = content_table_iterator_init(&tit, &subpage->tables);
+		while (1)
+		{
+			double y_paragraph;
+			double y_table;
+			block_t *block = (content && content->type == content_block) ? (block_t *)content : NULL;
+			paragraph_t *paragraph = (content && content->type == content_paragraph) ? (paragraph_t *)content : (block ? content_first_paragraph(&block->content) : NULL);
+			line_t *first_line = paragraph ? content_first_line(&paragraph->content) : NULL;
+			span_t *first_span = first_line ? content_first_span(&first_line->content) : NULL;
+			if (!paragraph && !table)   break;
+			y_paragraph = (first_span) ? first_span->chars[0].y : DBL_MAX;
+			y_table = (table) ? table->pos.y : DBL_MAX;
+
+			if (first_span && y_paragraph < y_table)
+			{
+				const matrix4_t *ctm = &first_span->ctm;
+				double           rotate = atan2(ctm->b, ctm->a);
+
+				if (spacing
+					&& content_state.ctm_prev
+					&& first_span
+					&& extract_matrix4_cmp(content_state.ctm_prev,
+								&first_span->ctm)
+					)
+				{
+					/* Extra vertical space between paragraphs that were at
+					different angles in the original document. */
+					if (odt_append_empty_paragraph(alloc, output, styles)) goto end;
+				}
+
+				if (spacing)
+				{
+					/* Extra vertical space between paragraphs. */
+					if (odt_append_empty_paragraph(alloc, output, styles)) goto end;
+				}
+
+				if (rotation && rotate != 0)
+				{
+					assert(block);
+					if (odt_append_rotated_paragraphs(alloc, &content_state, block, &text_box_id, ctm, rotate, output, styles)) goto end;
+				}
+				else if (block)
+				{
+					content_paragraph_iterator pit;
+					int                        first = 1;
+
+					for (paragraph = content_paragraph_iterator_init(&pit, &block->content); paragraph != NULL; paragraph = content_paragraph_iterator_next(&pit))
+					{
+						if (spacing && !first)
+						{
+							/* Extra vertical space between paragraphs. */
+							if (odt_append_empty_paragraph(alloc, output, styles)) goto end;
+						}
+						first = 0;
+
+						if (document_to_odt_content_paragraph(alloc, &content_state, paragraph, output, styles)) goto end;
+					}
+				}
+				else
+				{
+					if (document_to_odt_content_paragraph(alloc, &content_state, paragraph, output, styles)) goto end;
+				}
+				content = content_iterator_next(&cit);
+			}
+			else if (table)
+			{
+				if (odt_append_table(alloc, table, output, styles)) goto end;
+				table = content_table_iterator_next(&tit);
+			}
+		}
+
+		outf("images=%i", images);
+		if (images)
+		{
+			content_t *images, *next;
+			outf("subpage->images_num=%i", content_count_images(&subpage->content));
+			for (images = subpage->content.base.next; images != &subpage->content.base; images = next)
+			{
+				image_t *image = (image_t *)images;
+				next = images->next;
+				if (images->type != content_image)
+					continue;
+				odt_append_image(alloc, output, image);
+			}
+		}
+	}
+	ret = 0;
+
+	end:
+
+	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;
+}
+
+int
+extract_odt_content_item(
+		extract_alloc_t      *alloc,
+		extract_astring_t    *contentss,
+		int                   contentss_num,
+		extract_odt_styles_t *styles,
+		images_t             *images,
+		const char           *name,
+		const char           *text,
+		char                **text2)
+{
+	int e = -1;
+	extract_astring_t   temp;
+	extract_astring_init(&temp);
+	*text2 = NULL;
+
+	(void) images;
+	if (0)
+	{}
+	else if (!strcmp(name, "content.xml"))
+	{
+		/* Insert paragraphs content. */
+		char* text_intermediate = NULL;
+		extract_astring_t   styles_definitions = {0};
+
+		/* Insert content before '</office:text>'. */
+		if (extract_content_insert(
+				alloc,
+				text,
+				NULL /*single*/,
+				NULL /*mid_begin_name*/,
+				"</office:text>" /*mid_end_name*/,
+				contentss,
+				contentss_num,
+				&text_intermediate
+				)) goto end;
+		outf("text_intermediate: %s", text_intermediate);
+
+		/* Convert <styles> to text. */
+		if (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. */
+		if (extract_astring_cat(alloc, &styles_definitions,
+				"\n"
+				"<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(
+				alloc,
+				text_intermediate,
+				"<office:automatic-styles/>" /*single*/,
+				NULL /*mid_begin_name*/,
+				NULL /*mid_end_name*/,
+				&styles_definitions,
+				1,
+				text2
+				);
+		outf("e=%i errno=%i", e, errno);
+		extract_free(alloc, &text_intermediate);
+		extract_astring_free(alloc, &styles_definitions);
+		outf("e=%i errno=%i", e, errno);
+		if (e) goto end;
+	}
+	else if (!strcmp(name, "META-INF/manifest.xml"))
+	{
+		/* Add images. */
+		int e = 0;
+		int i;
+		for (i=0; i<images->images_num; ++i)
+		{
+			image_t* image = images->images[i];
+			if (!e) e = extract_astring_catf(alloc, &temp, "<manifest:file-entry manifest:full-path=\"Pictures/%s\" manifest:media-type=\"image/%s\"/>\n",
+					image->name,
+					image->type
+					);
+		}
+		if (!e) e = extract_content_insert(
+				alloc,
+				text,
+				NULL /*single*/,
+				NULL /*mid_begin_name*/,
+				"</manifest:manifest>" /*mid_end_name*/,
+				&temp,
+				1,
+				text2
+				);
+		if (e) goto end;
+	}
+	else
+	{
+		*text2 = NULL;
+	}
+	e = 0;
+	end:
+	outf("e=%i errno=%i text2=%s", e, errno, text2 ? *text2 : "");
+	if (e)
+	{
+		/* We might have set <text2> to new content. */
+		extract_free(alloc, text2);
+		/* We might have used <temp> as a temporary buffer. */
+	}
+	extract_astring_free(alloc, &temp);
+	extract_astring_init(&temp);
+	return e;
+}
+
+
+
+int
+extract_odt_write_template(
+		extract_alloc_t      *alloc,
+		extract_astring_t    *contentss,
+		int                   contentss_num,
+		extract_odt_styles_t *styles,
+		images_t             *images,
+		const char           *path_template,
+		const char           *path_out,
+		int                   preserve_dir)
+{
+	int     e = -1;
+	int     i;
+	char*   path_tempdir = NULL;
+	char*   path = NULL;
+	char*   text = NULL;
+	char*   text2 = NULL;
+
+	assert(path_out);
+	assert(path_template);
+
+	if (extract_check_path_shell_safe(path_out))
+	{
+		outf("path_out is unsafe: %s", path_out);
+		goto end;
+	}
+
+	outf("images->images_num=%i", images->images_num);
+	if (extract_asprintf(alloc, &path_tempdir, "%s.dir", path_out) < 0) goto end;
+	if (extract_systemf(alloc, "rm -r '%s' 2>/dev/null", path_tempdir) < 0) goto end;
+
+	if (extract_mkdir(path_tempdir, 0777))
+	{
+		outf("Failed to create directory: %s", path_tempdir);
+		goto end;
+	}
+
+	outf("Unzipping template document '%s' to tempdir: %s",
+			path_template, path_tempdir);
+	if (extract_systemf(alloc, "unzip -q -d '%s' '%s'", path_tempdir, path_template))
+	{
+		outf("Failed to unzip %s into %s",
+				path_template, path_tempdir);
+		goto end;
+	}
+
+	/* 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[] =
+		{
+				"content.xml",
+				"META-INF/manifest.xml",
+		};
+		int names_num = sizeof(names) / sizeof(names[0]);
+		for (i=0; i<names_num; ++i)
+		{
+			const char* name = names[i];
+			extract_free(alloc, &path);
+			extract_free(alloc, &text);
+			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,
+					contentss,
+					contentss_num,
+					styles,
+					images,
+					name,
+					text,
+					&text2
+					))
+			{
+				outf("extract_odt_content_item() failed");
+				goto end;
+			}
+
+			outf("after extract_odt_content_item styles->styles_num=%i", styles->styles_num);
+
+			{
+				const char* text3 = (text2) ? text2 : text;
+				if (extract_write_all(text3, strlen(text3), path)) goto end;
+				outf("have written to path=%s", path);
+			}
+		}
+	}
+
+	/* Copy images into <path_tempdir>/Pictures/. */
+	extract_free(alloc, &path);
+	if (extract_asprintf(alloc, &path, "%s/Pictures", path_tempdir) < 0) goto end;
+	if (extract_mkdir(path, 0777))
+	{
+		outf("Failed to mkdir %s", path);
+		goto end;
+	}
+	for (i=0; i<images->images_num; ++i)
+	{
+		image_t* image = images->images[i];
+		extract_free(alloc, &path);
+		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, '/');
+		if (!path_out_leaf) path_out_leaf = path_out;
+		if (extract_systemf(alloc, "cd '%s' && zip -q -r -D '../%s' .", path_tempdir, path_out_leaf))
+		{
+			outf("Zip command failed to convert '%s' directory into output file: %s",
+					path_tempdir, path_out);
+			goto end;
+		}
+	}
+
+	if (!preserve_dir)
+	{
+		if (extract_remove_directory(alloc, path_tempdir)) goto end;
+	}
+
+	e = 0;
+
+	end:
+	outf("e=%i", e);
+	extract_free(alloc, &path_tempdir);
+	extract_free(alloc, &path);
+	extract_free(alloc, &text);
+	extract_free(alloc, &text2);
+
+	if (e)
+	{
+		outf("Failed to create %s", path_out);
+	}
+	return e;
+}