io-gncxml-v2.h File Reference

api for GnuCash version 2 XML-based file format More...

#include <glib.h>
#include "gnc-engine.h"

Go to the source code of this file.

Data Structures
struct	conv_type

Functions
gint	gnc_xml2_find_ambiguous (const gchar *filename, GList *encodings, GHashTable **unique, GHashTable **ambiguous, GList **impossible)
	Read a file as plain byte stream to find words that are not completely ASCII. More...
gboolean	gnc_xml2_parse_with_subst (GncXmlBackend *xml_be, QofBook *book, GHashTable *subst)
	Parse a file in push mode, but replace byte sequences in the file given a hash table of substitutions. More...

Detailed Description

api for GnuCash version 2 XML-based file format

Author

Initial code by James LewisMoss, 2001

Definition in file io-gncxml-v2.h.

Function Documentation

gnc_xml2_find_ambiguous()

gint gnc_xml2_find_ambiguous	(	const gchar *	filename,
		GList *	encodings,
		GHashTable **	unique,
		GHashTable **	ambiguous,
		GList **	impossible
	)

Read a file as plain byte stream to find words that are not completely ASCII.

On error, unique, ambiguous and impossible will be filled up to that point, -1 will be returned.

Parameters

filename	Name of the file to read.
encodings	List of encodings to check words for, each begin one a GQuark in a pointer.
unique	Location used for a hash table for unique solutions, if not NULL. The byte sequence is the key, successful_conversion the value.
ambiguous	Location used for a hash table for ambiguous byte sequences, if not NULL. The byte sequences is the key, a list of successful_conversions the value.
impossible	Location used for a list for undecodable byte sequences, if not NULL.

Returns

Size of impossible, -1 on error.

Definition at line 1790 of file io-gncxml-v2.cpp.

{
    GList* iconv_list = NULL, *conv_list = NULL, *iter;
    iconv_item_type* iconv_item = NULL, *ascii = NULL;
    const gchar* enc;
    GHashTable* processed = NULL;
    gint n_impossible = 0;
    GError* error = NULL;
    gboolean clean_return = FALSE;

    auto [file, thread] = try_gz_open (filename, "r",
                                       is_gzipped_file (filename), FALSE);
    if (file == NULL)
    {
        PWARN ("Unable to open file %s", filename);
        goto cleanup_find_ambs;
    }

    /* we need ascii */
    ascii = g_new (iconv_item_type, 1);
    ascii->encoding = g_quark_from_string ("ASCII");
    ascii->iconv = g_iconv_open ("UTF-8", "ASCII");
    if (ascii->iconv == (GIConv) - 1)
    {
        PWARN ("Unable to open ASCII ICONV conversion descriptor");
        goto cleanup_find_ambs;
    }

    /* call iconv_open on encodings */
    for (iter = encodings; iter; iter = iter->next)
    {
        iconv_item = g_new (iconv_item_type, 1);
        iconv_item->encoding = GPOINTER_TO_UINT (iter->data);
        if (iconv_item->encoding == ascii->encoding)
        {
            continue;
        }

        enc = g_quark_to_string (iconv_item->encoding);
        iconv_item->iconv = g_iconv_open ("UTF-8", enc);
        if (iconv_item->iconv == (GIConv) - 1)
        {
            PWARN ("Unable to open IConv conversion descriptor for '%s'", enc);
            g_free (iconv_item);
            goto cleanup_find_ambs;
        }
        else
        {
            iconv_list = g_list_prepend (iconv_list, iconv_item);
        }
    }

    /* prepare data containers */
    if (unique)
        *unique = g_hash_table_new_full (g_str_hash, g_str_equal, g_free,
                                         (GDestroyNotify) conv_free);
    if (ambiguous)
        *ambiguous = g_hash_table_new_full (g_str_hash, g_str_equal, g_free,
                                            (GDestroyNotify) conv_list_free);
    if (impossible)
        *impossible = NULL;
    processed = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, NULL);

    /* loop through lines */
    while (1)
    {
        gchar line[256], *word, *utf8;
        gchar** word_array, **word_cursor;
        conv_type* conv = NULL;

        if (!fgets (line, sizeof (line) - 1, file))
        {
            if (feof (file))
            {
                break;
            }
            else
            {
                goto cleanup_find_ambs;
            }
        }

        g_strchomp (line);
        replace_character_references (line);
        word_array = g_strsplit_set (line, "> <", 0);

        /* loop through words */
        for (word_cursor = word_array; *word_cursor; word_cursor++)
        {
            word = *word_cursor;
            if (!word)
                continue;

            utf8 = g_convert_with_iconv (word, -1, ascii->iconv,
                                         NULL, NULL, &error);
            if (utf8)
            {
                /* pure ascii */
                g_free (utf8);
                continue;
            }
            g_error_free (error);
            error = NULL;

            if (g_hash_table_lookup_extended (processed, word, NULL, NULL))
            {
                /* already processed */
                continue;
            }

            /* loop through encodings */
            conv_list = NULL;
            for (iter = iconv_list; iter; iter = iter->next)
            {
                iconv_item = static_cast<decltype (iconv_item)> (iter->data);
                utf8 = g_convert_with_iconv (word, -1, iconv_item->iconv,
                                             NULL, NULL, &error);
                if (utf8)
                {
                    conv = g_new (conv_type, 1);
                    conv->encoding = iconv_item->encoding;
                    conv->utf8_string = utf8;
                    conv_list = g_list_prepend (conv_list, conv);
                }
                else
                {
                    g_error_free (error);
                    error = NULL;
                }
            }

            /* no successful conversion */
            if (!conv_list)
            {
                if (impossible)
                    *impossible = g_list_append (*impossible, g_strdup (word));
                n_impossible++;
            }

            /* more than one successful conversion */
            else if (conv_list->next)
            {
                if (ambiguous)
                {
                    g_hash_table_insert (*ambiguous, g_strdup (word), conv_list);
                }
                else
                {
                    conv_list_free (conv_list);
                }
            }

            /* only one successful conversion */
            else
            {
                if (unique)
                {
                    g_hash_table_insert (*unique, g_strdup (word), conv);
                }
                else
                {
                    conv_free (conv);
                }
                g_list_free (conv_list);
            }

            g_hash_table_insert (processed, g_strdup (word), NULL);
        }
        g_strfreev (word_array);
    }

    clean_return = TRUE;

cleanup_find_ambs:

    if (iconv_list)
    {
        for (iter = iconv_list; iter; iter = iter->next)
        {
            if (iter->data)
            {
                g_iconv_close (((iconv_item_type*) iter->data)->iconv);
                g_free (iter->data);
            }
        }
        g_list_free (iconv_list);
    }
    if (processed)
        g_hash_table_destroy (processed);
    if (ascii)
        g_free (ascii);
    if (file)
    {
        fclose (file);
        if (thread)
            g_thread_join (thread);
    }

    return (clean_return) ? n_impossible : -1;
}

gnc_xml2_parse_with_subst()

gboolean gnc_xml2_parse_with_subst	(	GncXmlBackend *	xml_be,
		QofBook *	book,
		GHashTable *	subst
	)

Parse a file in push mode, but replace byte sequences in the file given a hash table of substitutions.

Parameters

subst

hash table with keys and values of type gchar*

Definition at line 2129 of file io-gncxml-v2.cpp.

{
    push_data_type* push_data;
    gboolean success;

    push_data = g_new (push_data_type, 1);
    push_data->filename = xml_be->get_filename();
    push_data->subst = subst;

    success = qof_session_load_from_xml_file_v2_full (
                  xml_be, book, (sixtp_push_handler) parse_with_subst_push_handler,
                  push_data, GNC_BOOK_XML2_FILE);
    g_free (push_data);

    if (success)
        qof_instance_set_dirty (QOF_INSTANCE (book));

    return success;
}