Initial working rich strings.

Issue #37

.indent.pro

@@ -113,6 +113,7 @@
 -T lxw_relationships
 -T lxw_repeat_cols
 -T lxw_repeat_rows
+-T lxw_rich_string_tuple
 -T lxw_row
 -T lxw_row_col_options
 -T lxw_row_t

include/xlsxwriter/shared_strings.h

@@ -37,6 +37,7 @@ STAILQ_HEAD(sst_order_list, sst_element);
 struct sst_element {
     uint32_t index;
     char *string;
+    uint8_t is_rich_string;
 
     STAILQ_ENTRY (sst_element) sst_order_pointers;
     RB_ENTRY (sst_element) sst_tree_pointers;
@@ -64,7 +65,8 @@ extern "C" {
 
 lxw_sst *lxw_sst_new();
 void lxw_sst_free(lxw_sst *sst);
-struct sst_element *lxw_get_sst_index(lxw_sst *sst, const char *string);
+struct sst_element *lxw_get_sst_index(lxw_sst *sst, const char *string,
+                                      uint8_t is_rich_string);
 void lxw_sst_assemble_xml_file(lxw_sst *self);
 
 /* Declarations required for unit testing. */

include/xlsxwriter/styles.h

@@ -10,6 +10,7 @@
 #define __LXW_STYLES_H__
 
 #include <stdint.h>
+#include <ctype.h>
 
 #include "format.h"
 
@@ -40,6 +41,8 @@ extern "C" {
 lxw_styles *lxw_styles_new();
 void lxw_styles_free(lxw_styles *styles);
 void lxw_styles_assemble_xml_file(lxw_styles *self);
+void lxw_styles_write_string_fragment(lxw_styles *self, char *string);
+void lxw_styles_write_rich_font(lxw_styles *lxw_styles, lxw_format *format);
 
 /* Declarations required for unit testing. */
 #ifdef TESTING
@@ -48,10 +51,12 @@ STATIC void _styles_xml_declaration(lxw_styles *self);
 STATIC void _write_style_sheet(lxw_styles *self);
 STATIC void _write_font_size(lxw_styles *self, double font_size);
 STATIC void _write_font_color_theme(lxw_styles *self, uint8_t theme);
-STATIC void _write_font_name(lxw_styles *self, const char *font_name);
+STATIC void _write_font_name(lxw_styles *self, const char *font_name,
+                             uint8_t is_rich_string);
 STATIC void _write_font_family(lxw_styles *self, uint8_t font_family);
 STATIC void _write_font_scheme(lxw_styles *self, const char *font_scheme);
-STATIC void _write_font(lxw_styles *self, lxw_format *format);
+STATIC void _write_font(lxw_styles *self, lxw_format *format,
+                        uint8_t is_rich_string);
 STATIC void _write_fonts(lxw_styles *self);
 STATIC void _write_default_fill(lxw_styles *self, const char *pattern);
 STATIC void _write_fills(lxw_styles *self);

include/xlsxwriter/worksheet.h

@@ -53,6 +53,7 @@
 #include "drawing.h"
 #include "common.h"
 #include "format.h"
+#include "styles.h"
 #include "utility.h"
 
 #define LXW_ROW_MAX           1048576
@@ -640,6 +641,11 @@ typedef struct lxw_protection {
     char hash[5];
 } lxw_protection;
 
+typedef struct lxw_rich_string_tuple {
+    lxw_format *format;
+    char *string;
+} lxw_rich_string_tuple;
+
 /**
  * @brief Struct to represent an Excel worksheet.
  *
@@ -1314,6 +1320,12 @@ lxw_error worksheet_write_formula_num(lxw_worksheet *worksheet,
                                       const char *formula,
                                       lxw_format *format, double result);
 
+lxw_error worksheet_write_rich_string(lxw_worksheet *worksheet,
+                                      lxw_row_t row_num,
+                                      lxw_col_t col_num,
+                                      lxw_rich_string_tuple *rich_strings[],
+                                      lxw_format *format);
+
 /**
  * @brief Set the properties for a row of cells.
  *

include/xlsxwriter/xmlwriter.h

@@ -165,6 +165,8 @@ void lxw_xml_data_element(FILE * xmlfile,
                           const char *data,
                           struct xml_attribute_list *attributes);
 
+void lxw_xml_rich_si_element(FILE * xmlfile, const char *string);
+
 char *lxw_escape_control_characters(const char *string);
 
 char *lxw_escape_data(const char *data);

src/shared_strings.c

@@ -161,6 +161,15 @@ _write_si(lxw_sst *self, char *string)
         free(string);
 }
 
+/*
+ * Write the <si> element for rich strings.
+ */
+STATIC void
+_write_rich_si(lxw_sst *self, char *string)
+{
+    lxw_xml_rich_si_element(self->file, string);
+}
+
 /*
  * Write the <sst> element.
  */
@@ -198,7 +207,11 @@ _write_sst_strings(lxw_sst *self)
 
     STAILQ_FOREACH(sst_element, self->order_list, sst_order_pointers) {
         /* Write the si element. */
+        if (sst_element->is_rich_string)
+            _write_rich_si(self, sst_element->string);
+        else
             _write_si(self, sst_element->string);
+
     }
 }
 
@@ -230,7 +243,7 @@ lxw_sst_assemble_xml_file(lxw_sst *self)
  * Add to or find a string in the SST SharedString table and return it's index.
  */
 struct sst_element *
-lxw_get_sst_index(lxw_sst *sst, const char *string)
+lxw_get_sst_index(lxw_sst *sst, const char *string, uint8_t is_rich_string)
 {
     struct sst_element *element;
     struct sst_element *existing_element;
@@ -243,6 +256,7 @@ lxw_get_sst_index(lxw_sst *sst, const char *string)
     /* Create potential new element with the string and its index. */
     element->index = sst->unique_count;
     element->string = lxw_strdup(string);
+    element->is_rich_string = is_rich_string;
 
     /* Try to insert it and see whether we already have that string. */
     existing_element = RB_INSERT(sst_rb_tree, sst->rb_tree, element);

src/styles.c

@@ -14,6 +14,8 @@
 /*
  * Forward declarations.
  */
+STATIC void _write_font(lxw_styles *self, lxw_format *format,
+                        uint8_t is_rich_string);
 
 /*****************************************************************************
  *
@@ -66,6 +68,34 @@ lxw_styles_free(lxw_styles *styles)
     free(styles);
 }
 
+/*
+ * Write the <t> element for rich strings.
+ */
+void
+lxw_styles_write_string_fragment(lxw_styles *self, char *string)
+{
+    struct xml_attribute_list attributes;
+    struct xml_attribute *attribute;
+
+    LXW_INIT_ATTRIBUTES();
+
+    /* Add attribute to preserve leading or trailing whitespace. */
+    if (isspace((unsigned char) string[0])
+        || isspace((unsigned char) string[strlen(string) - 1]))
+        LXW_PUSH_ATTRIBUTES_STR("xml:space", "preserve");
+
+    lxw_xml_data_element(self->file, "t", string, &attributes);
+
+    LXW_FREE_ATTRIBUTES();
+}
+
+void
+lxw_styles_write_rich_font(lxw_styles *self, lxw_format *format)
+{
+
+    _write_font(self, format, LXW_TRUE);
+}
+
 /*****************************************************************************
  *
  * XML functions.
@@ -207,7 +237,8 @@ _write_font_color_rgb(lxw_styles *self, int32_t rgb)
  * Write the <name> element.
  */
 STATIC void
-_write_font_name(lxw_styles *self, const char *font_name)
+_write_font_name(lxw_styles *self, const char *font_name,
+                 uint8_t is_rich_string)
 {
     struct xml_attribute_list attributes;
     struct xml_attribute *attribute;
@@ -219,6 +250,9 @@ _write_font_name(lxw_styles *self, const char *font_name)
     else
         LXW_PUSH_ATTRIBUTES_STR("val", LXW_DEFAULT_FONT_NAME);
 
+    if (is_rich_string)
+        lxw_xml_empty_tag(self->file, "rFont", &attributes);
+    else
         lxw_xml_empty_tag(self->file, "name", &attributes);
 
     LXW_FREE_ATTRIBUTES();
@@ -309,8 +343,11 @@ _write_vert_align(lxw_styles *self, const char *align)
  * Write the <font> element.
  */
 STATIC void
-_write_font(lxw_styles *self, lxw_format *format)
+_write_font(lxw_styles *self, lxw_format *format, uint8_t is_rich_string)
 {
+    if (is_rich_string)
+        lxw_xml_start_tag(self->file, "rPr", NULL);
+    else
         lxw_xml_start_tag(self->file, "font", NULL);
 
     if (format->bold)
@@ -347,7 +384,7 @@ _write_font(lxw_styles *self, lxw_format *format)
     else
         _write_font_color_theme(self, LXW_DEFAULT_FONT_THEME);
 
-    _write_font_name(self, format->font_name);
+    _write_font_name(self, format->font_name, is_rich_string);
     _write_font_family(self, format->font_family);
 
     /* Only write the scheme element for the default font type if it
@@ -358,6 +395,9 @@ _write_font(lxw_styles *self, lxw_format *format)
         _write_font_scheme(self, format->font_scheme);
     }
 
+    if (is_rich_string)
+        lxw_xml_end_tag(self->file, "rPr");
+    else
         lxw_xml_end_tag(self->file, "font");
 }
 
@@ -378,7 +418,7 @@ _write_fonts(lxw_styles *self)
 
     STAILQ_FOREACH(format, self->xf_formats, list_pointers) {
         if (format->has_font)
-            _write_font(self, format);
+            _write_font(self, format, LXW_FALSE);
     }
 
     lxw_xml_end_tag(self->file, "fonts");

src/worksheet.c

@@ -3842,7 +3842,7 @@ worksheet_write_string(lxw_worksheet *self,
 
     if (!self->optimize) {
         /* Get the SST element and string id. */
-        sst_element = lxw_get_sst_index(self->sst, string);
+        sst_element = lxw_get_sst_index(self->sst, string, LXW_FALSE);
 
         if (!sst_element)
             return LXW_ERROR_SHARED_STRING_INDEX_NOT_FOUND;
@@ -4317,6 +4317,123 @@ worksheet_write_url(lxw_worksheet *self,
                                    NULL);
 }
 
+/*
+ * Write a rich string to an Excel file.
+ *
+ * Rather than duplicate several of the styles.c font xml methods of styles.c
+ * and write the data to a memory buffer this function creates a temporary
+ * styles object and uses it to write the data to a file. It then reads that
+ * data back into memory and closes the file.
+ */
+lxw_error
+worksheet_write_rich_string(lxw_worksheet *self,
+                            lxw_row_t row_num,
+                            lxw_col_t col_num,
+                            lxw_rich_string_tuple *rich_strings[],
+                            lxw_format *format)
+{
+    lxw_cell *cell;
+    int32_t string_id;
+    struct sst_element *sst_element;
+    lxw_error err;
+    uint8_t i;
+    long file_size;
+    char *rich_string = NULL;
+    lxw_styles *styles = NULL;
+    lxw_format *default_format = NULL;
+    lxw_rich_string_tuple *rich_string_tuple = NULL;
+    FILE *tmpfile;
+
+    err = _check_dimensions(self, row_num, col_num, LXW_FALSE, LXW_FALSE);
+    if (err)
+        return err;
+
+    /* Create a tmp file for the styles object. */
+    tmpfile = lxw_tmpfile(self->tmpdir);
+    if (!tmpfile)
+        return LXW_ERROR_CREATING_TMPFILE;
+
+    /* Create a temp styles object for writing the font data. */
+    styles = lxw_styles_new();
+    GOTO_LABEL_ON_MEM_ERROR(styles, mem_error);
+    styles->file = tmpfile;
+
+    /* Create a default format for non-formatted text. */
+    default_format = lxw_format_new();
+    GOTO_LABEL_ON_MEM_ERROR(default_format, mem_error);
+
+    /* Iterate through the rich string fragments and write each one out. */
+    i = 0;
+    while ((rich_string_tuple = rich_strings[i++]) != NULL) {
+        lxw_xml_start_tag(tmpfile, "r", NULL);
+
+        if (rich_string_tuple->format) {
+            /* Write the user defined font format. */
+            lxw_styles_write_rich_font(styles, rich_string_tuple->format);
+        }
+        else {
+            /* Write a default font format. Except for the first fragment. */
+            if (i > 1)
+                lxw_styles_write_rich_font(styles, default_format);
+        }
+
+        lxw_styles_write_string_fragment(styles, rich_string_tuple->string);
+        lxw_xml_end_tag(tmpfile, "r");
+    }
+
+    /* Free the temp objects. */
+    lxw_styles_free(styles);
+    lxw_format_free(default_format);
+
+    /* Flush the file and read the size to calculate the required memory. */
+    fflush(tmpfile);
+    file_size = ftell(tmpfile);
+
+    /* Allocate a buffer for the rich string xml data. */
+    rich_string = calloc(file_size + 1, 1);
+    GOTO_LABEL_ON_MEM_ERROR(rich_string, mem_error);
+
+    /* Rewind the file and read the data into the memory buffer. */
+    rewind(tmpfile);
+    if (fread(rich_string, file_size, 1, tmpfile) < 1) {
+         /* TODO */
+         fclose(tmpfile);
+        return LXW_ERROR_MAX_STRING_LENGTH_EXCEEDED;
+    }
+
+    /* Close the temp file. */
+    fclose(tmpfile);
+
+    if (lxw_utf8_strlen(rich_string) > LXW_STR_MAX) {
+        free(rich_string);
+        return LXW_ERROR_MAX_STRING_LENGTH_EXCEEDED;
+    }
+
+    if (!self->optimize) {
+        /* Get the SST element and string id. */
+        sst_element = lxw_get_sst_index(self->sst, rich_string, LXW_TRUE);
+        free(rich_string);
+
+        if (!sst_element)
+            return LXW_ERROR_SHARED_STRING_INDEX_NOT_FOUND;
+
+        string_id = sst_element->index;
+        cell = _new_string_cell(row_num, col_num, string_id,
+                                sst_element->string, format);
+
+        _insert_cell(self, row_num, col_num, cell);
+    }
+
+    return LXW_NO_ERROR;
+
+mem_error:
+    lxw_styles_free(styles);
+    lxw_format_free(default_format);
+    fclose(tmpfile);
+
+    return LXW_ERROR_MEMORY_MALLOC_FAILED;
+}
+
 /*
  * Set the properties of a single column or a range of columns with options.
  */

src/xmlwriter.c

@@ -140,6 +140,15 @@ lxw_xml_data_element(FILE * xmlfile,
     fprintf(xmlfile, "</%s>", tag);
 }
 
+/*
+ * Write an XML <si> element for rich strings, without encoding.
+ */
+void
+lxw_xml_rich_si_element(FILE * xmlfile, const char *string)
+{
+    fprintf(xmlfile, "<si>%s</si>", string);
+}
+
 /*
  * Escape XML characters in attributes.
  */

test/functional/src/test_rich_string01.c

@@ -0,0 +1,34 @@
+/*****************************************************************************
+ * Test cases for libxlsxwriter.
+ *
+ * Test to compare output against Excel files.
+ *
+ * Copyright 2014-2018, John McNamara, jmcnamara@cpan.org
+ *
+ */
+
+#include "xlsxwriter.h"
+
+int main() {
+
+    lxw_workbook  *workbook  = new_workbook("test_rich_string01.xlsx");
+    lxw_worksheet *worksheet = workbook_add_worksheet(workbook, NULL);
+
+    lxw_format *bold = workbook_add_format(workbook);
+    lxw_format *italic = workbook_add_format(workbook);
+
+    format_set_bold(bold);
+    format_set_italic(italic);
+
+    worksheet_write_string(worksheet, CELL("A1"), "Foo", bold);
+    worksheet_write_string(worksheet, CELL("A2"), "Bar", italic);
+
+    lxw_rich_string_tuple fragment1 = {.format = NULL, .string = "a"};
+    lxw_rich_string_tuple fragment2 = {.format = bold, .string = "bc"};
+    lxw_rich_string_tuple fragment3 = {.format = NULL, .string = "defg"};
+
+    lxw_rich_string_tuple *rich_strings[] = {&fragment1, &fragment2, &fragment3, NULL};
+    worksheet_write_rich_string(worksheet, CELL("A3"), rich_strings, NULL);
+
+    return workbook_close(workbook);
+}

test/functional/test_rich_string.py

@@ -0,0 +1,17 @@
+###############################################################################
+#
+# Tests for libxlsxwriter.
+#
+# Copyright 2014-2018, John McNamara, jmcnamara@cpan.org
+#
+
+import base_test_class
+
+class TestCompareXLSXFiles(base_test_class.XLSXBaseTest):
+    """
+    Test file created with libxlsxwriter against a file created by Excel.
+
+    """
+
+    def test_rich_string01(self):
+        self.run_exe_test('test_rich_string01')

test/unit/sst/test_shared_strings.c

@@ -33,13 +33,13 @@ CTEST(sst, sst01) {
     lxw_sst *sst = lxw_sst_new();
     sst->file = testfile;
 
-    lxw_get_sst_index(sst, "neptune");
-    lxw_get_sst_index(sst, "neptune");
-    lxw_get_sst_index(sst, "neptune");
-    lxw_get_sst_index(sst, "mars");
-    lxw_get_sst_index(sst, "mars");
-    lxw_get_sst_index(sst, "venus");
-    lxw_get_sst_index(sst, "venus");
+    lxw_get_sst_index(sst, "neptune", LXW_FALSE);
+    lxw_get_sst_index(sst, "neptune", LXW_FALSE);
+    lxw_get_sst_index(sst, "neptune", LXW_FALSE);
+    lxw_get_sst_index(sst, "mars", LXW_FALSE);
+    lxw_get_sst_index(sst, "mars", LXW_FALSE);
+    lxw_get_sst_index(sst, "venus", LXW_FALSE);
+    lxw_get_sst_index(sst, "venus", LXW_FALSE);
 
     lxw_sst_assemble_xml_file(sst);
 
@@ -72,9 +72,9 @@ CTEST(sst, sst02) {
     sst->file = testfile;
 
     // Test strings with whitespace that must be preserved.
-    lxw_get_sst_index(sst, "abcdefg");
-    lxw_get_sst_index(sst, "   abcdefg");
-    lxw_get_sst_index(sst, "abcdefg   ");
+    lxw_get_sst_index(sst, "abcdefg", LXW_FALSE);
+    lxw_get_sst_index(sst, "   abcdefg", LXW_FALSE);
+    lxw_get_sst_index(sst, "abcdefg   ", LXW_FALSE);
 
     lxw_sst_assemble_xml_file(sst);
 
@@ -82,4 +82,3 @@ CTEST(sst, sst02) {
 
     lxw_sst_free(sst);
 }
-

test/unit/styles/test_styles_write_font.c

@@ -23,7 +23,7 @@ CTEST(styles, write_font01) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -46,7 +46,7 @@ CTEST(styles, write_font02) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -69,7 +69,7 @@ CTEST(styles, write_font03) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -92,7 +92,7 @@ CTEST(styles, write_font04) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -115,7 +115,7 @@ CTEST(styles, write_font05) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -138,7 +138,7 @@ CTEST(styles, write_font06) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -161,7 +161,7 @@ CTEST(styles, write_font07) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -184,7 +184,7 @@ CTEST(styles, write_font08) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -207,7 +207,7 @@ CTEST(styles, write_font09) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -230,7 +230,7 @@ CTEST(styles, write_font10) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -253,7 +253,7 @@ CTEST(styles, write_font11) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -276,7 +276,7 @@ CTEST(styles, write_font12) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -307,7 +307,7 @@ CTEST(styles, write_font13) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -330,7 +330,7 @@ CTEST(styles, write_font14) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -353,7 +353,7 @@ CTEST(styles, write_font15) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -376,7 +376,7 @@ CTEST(styles, write_font16) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
@@ -403,13 +403,10 @@ CTEST(styles, write_font17) {
 
     styles->file = testfile;
 
-    _write_font(styles, format);
+    _write_font(styles, format, LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);
 
     lxw_styles_free(styles);
     lxw_format_free(format);
 }
-
-
-

test/unit/styles/test_styles_write_font_name.c

@@ -21,7 +21,7 @@ CTEST(styles, write_name) {
     lxw_styles *styles = lxw_styles_new();
     styles->file = testfile;
 
-    _write_font_name(styles, "Calibri");
+    _write_font_name(styles, "Calibri", LXW_FALSE);
 
     RUN_XLSX_STREQ(exp, got);