* src/import-export/import-parse.[ch]: routines to parse numbers

and dates in a mostly-locale-agnostic way. * src/import-export/import-format-dialog.c: a dialog to ask the user to choose a format, when parsing just isn't enough. * src/import-export/generic-import.glade: add the format_match dialog * src/import-export/test/test-import-parse.c: test the import-parse routines. git-svn-id: svn+ssh://svn.gnucash.org/repo/gnucash/trunk@8851 57a11ea4-9604-0410-9ed3-97b8803252fd
2025-02-25 18:55:30 -06:00 · 2003-07-11 03:28:00 +00:00 · 2003-07-11 03:28:00 +00:00 · 06aa3a6845
commit 06aa3a6845
parent 7b3bb91ed7
8 changed files with 901 additions and 14 deletions
--- a/10
+++ b/10
@ -1,3 +1,13 @@
+2003-07-10  Derek Atkins  <derek@ihtfp.com>
+
+	* src/import-export/import-parse.[ch]: routines to parse numbers
+	  and dates in a mostly-locale-agnostic way.
+	* src/import-export/import-format-dialog.c: a dialog to ask the user
+	  to choose a format, when parsing just isn't enough.
+	* src/import-export/generic-import.glade: add the format_match dialog
+	* src/import-export/test/test-import-parse.c: test the import-parse
+	  routines.
+
 2003-07-09  Derek Atkins  <derek@ihtfp.com>

 	* configure.in: change (and comment out) checks for pthreads
--- a/src/import-export/Makefile.am
+++ b/src/import-export/Makefile.am
@ -6,7 +6,9 @@ libgncmod_generic_import_la_SOURCES = \
 	import-account-matcher.c \
 	import-commodity-matcher.c \
 	import-backend.c \
+	import-format-dialog.c \
 	import-match-picker.c \
+	import-parse.c \
 	import-utilities.c \
 	import-settings.c \
 	import-match-map.c \
@ -15,7 +17,8 @@ libgncmod_generic_import_la_SOURCES = \

 gncincludedir = ${GNC_INCLUDE_DIR}
 gncinclude_HEADERS = \
-  import-match-map.h
+  import-match-map.h \
+  import-parse.h

 noinst_HEADERS = \
  import-account-matcher.h \
@ -68,16 +71,24 @@ EXTRA_DIST = \
  .cvsignore \
  generic-import-design.txt

-#noinst_DATA = .scm-links
+if GNUCASH_SEPARATE_BUILDDIR
+#For compiling
+SCM_FILE_LINKS = 
+#For executing test cases
+SCM_FILE_LINKS += generic-import.scm
+endif

-#.scm-links:
-#	rm -f g-wrapped gnucash import-export qif-import
-#	ln -sf . qif-import # to fix (load "qif-import/foo.scm")
-#	ln -sf . import-export
-#	ln -sf . gnucash
-#	ln -sf . g-wrapped
-#	touch .scm-links
-DISTCLEANFILES = gnucash g-wrapped .scm-links import-export
+noinst_DATA = .scm-links
+
+.scm-links:
+	rm -f g-wrapped gnucash generic-import import-export
+	ln -sf . import-export
+	ln -sf . gnucash
+	ln -sf . generic-import
+	ln -sf . g-wrapped
+	touch .scm-links
+
+DISTCLEANFILES = gnucash generic-import g-wrapped .scm-links import-export

 # FIXME remove this when qif-io-core is finished
 DIST_SUBDIRS = binary-import qif-import qif-io-core ofx test hbci log-replay
--- a/src/import-export/generic-import.glade
+++ b/src/import-export/generic-import.glade
@ -1092,4 +1092,140 @@ click &quot;Ok&quot;.</label>
  </widget>
 </widget>

+<widget>
+  <class>GnomeDialog</class>
+  <name>format_picker</name>
+  <title>Choose a format</title>
+  <type>GTK_WINDOW_TOPLEVEL</type>
+  <position>GTK_WIN_POS_CENTER</position>
+  <modal>True</modal>
+  <allow_shrink>False</allow_shrink>
+  <allow_grow>False</allow_grow>
+  <auto_shrink>False</auto_shrink>
+  <auto_close>False</auto_close>
+  <hide_on_close>False</hide_on_close>
+
+  <widget>
+    <class>GtkVBox</class>
+    <child_name>GnomeDialog:vbox</child_name>
+    <name>dialog-vbox4</name>
+    <homogeneous>False</homogeneous>
+    <spacing>8</spacing>
+    <child>
+      <padding>4</padding>
+      <expand>True</expand>
+      <fill>True</fill>
+    </child>
+
+    <widget>
+      <class>GtkHButtonBox</class>
+      <child_name>GnomeDialog:action_area</child_name>
+      <name>dialog-action_area5</name>
+      <layout_style>GTK_BUTTONBOX_END</layout_style>
+      <spacing>8</spacing>
+      <child_min_width>85</child_min_width>
+      <child_min_height>27</child_min_height>
+      <child_ipad_x>7</child_ipad_x>
+      <child_ipad_y>0</child_ipad_y>
+      <child>
+	<padding>0</padding>
+	<expand>False</expand>
+	<fill>True</fill>
+	<pack>GTK_PACK_END</pack>
+      </child>
+
+      <widget>
+	<class>GtkButton</class>
+	<name>button79</name>
+	<can_default>True</can_default>
+	<can_focus>True</can_focus>
+	<stock_button>GNOME_STOCK_BUTTON_OK</stock_button>
+      </widget>
+    </widget>
+
+    <widget>
+      <class>GtkFrame</class>
+      <name>frame2</name>
+      <label>Choose a format</label>
+      <label_xalign>0</label_xalign>
+      <shadow_type>GTK_SHADOW_ETCHED_IN</shadow_type>
+      <child>
+	<padding>0</padding>
+	<expand>True</expand>
+	<fill>True</fill>
+      </child>
+
+      <widget>
+	<class>GtkVBox</class>
+	<name>vbox8</name>
+	<border_width>5</border_width>
+	<homogeneous>False</homogeneous>
+	<spacing>15</spacing>
+
+	<widget>
+	  <class>GtkLabel</class>
+	  <name>msg_label</name>
+	  <label>do not translate</label>
+	  <justify>GTK_JUSTIFY_CENTER</justify>
+	  <wrap>True</wrap>
+	  <xalign>0.5</xalign>
+	  <yalign>0.5</yalign>
+	  <xpad>0</xpad>
+	  <ypad>0</ypad>
+	  <child>
+	    <padding>0</padding>
+	    <expand>False</expand>
+	    <fill>False</fill>
+	  </child>
+	</widget>
+
+	<widget>
+	  <class>GtkHBox</class>
+	  <name>hbox1</name>
+	  <homogeneous>False</homogeneous>
+	  <spacing>3</spacing>
+	  <child>
+	    <padding>0</padding>
+	    <expand>True</expand>
+	    <fill>True</fill>
+	  </child>
+
+	  <widget>
+	    <class>GtkLabel</class>
+	    <name>label847791</name>
+	    <label>Format:</label>
+	    <justify>GTK_JUSTIFY_LEFT</justify>
+	    <wrap>False</wrap>
+	    <xalign>1</xalign>
+	    <yalign>0.5</yalign>
+	    <xpad>0</xpad>
+	    <ypad>0</ypad>
+	    <child>
+	      <padding>0</padding>
+	      <expand>False</expand>
+	      <fill>False</fill>
+	    </child>
+	  </widget>
+
+	  <widget>
+	    <class>GtkHBox</class>
+	    <name>menu_box</name>
+	    <homogeneous>False</homogeneous>
+	    <spacing>0</spacing>
+	    <child>
+	      <padding>0</padding>
+	      <expand>True</expand>
+	      <fill>True</fill>
+	    </child>
+
+	    <widget>
+	      <class>Placeholder</class>
+	    </widget>
+	  </widget>
+	</widget>
+      </widget>
+    </widget>
+  </widget>
+</widget>
+
 </GTK-Interface>
--- a/src/import-export/import-format-dialog.c
+++ b/src/import-export/import-format-dialog.c
@ -0,0 +1,119 @@
+/*
+ * import-format-dialog.c -- provides a UI to ask for users to resolve ambiguities.
+ *
+ * Created by:	Derek Atkins <derek@ihtfp.com>
+ *
+ */
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <glib.h>
+
+#include <glade/glade.h>
+
+#include "import-parse.h"
+#include "dialog-utils.h"
+#include "gnc-ui-util.h"
+
+#define MAX_CHOICES 6
+
+static void
+choice_option_changed (GtkWidget *widget, gint index, gpointer index_p)
+{
+  gint *my_index = index_p;
+  *my_index = index;
+}
+
+static GncImportFormat
+add_menu_and_run_dialog(GtkWidget *dialog, GtkWidget *menu_box, GncImportFormat fmt)
+{
+  GtkWidget *menu;
+  gint index = 0, count = 0;
+  GncImportFormat formats[MAX_CHOICES];
+  GNCOptionInfo menus[MAX_CHOICES];
+
+  memset(&menus, 0, sizeof(menus));
+
+  if (fmt & GNCIF_NUM_PERIOD) {
+    formats[count] = GNCIF_NUM_PERIOD;
+    menus[count].name = _("Period: 123,456.78");
+    menus[count].callback = choice_option_changed;
+    menus[count].user_data = &index;
+    count++;
+  }
+
+  if (fmt & GNCIF_NUM_COMMA) {
+    formats[count] = GNCIF_NUM_COMMA;
+    menus[count].name = _("Comma: 123.456,78");
+    menus[count].callback = choice_option_changed;
+    menus[count].user_data = &index;
+    count++;
+  }
+
+  if (fmt & GNCIF_DATE_MDY) {
+    formats[count] = GNCIF_DATE_MDY;
+    menus[count].name = _("m/d/y");
+    menus[count].callback = choice_option_changed;
+    menus[count].user_data = &index;
+    count++;
+  }
+
+  if (fmt & GNCIF_DATE_DMY) {
+    formats[count] = GNCIF_DATE_DMY;
+    menus[count].name = _("d/m/y");
+    menus[count].callback = choice_option_changed;
+    menus[count].user_data = &index;
+    count++;
+  }
+
+  if (fmt & GNCIF_DATE_YMD) {
+    formats[count] = GNCIF_DATE_YMD;
+    menus[count].name = _("y/m/d");
+    menus[count].callback = choice_option_changed;
+    menus[count].user_data = &index;
+    count++;
+  }
+
+  if (fmt & GNCIF_DATE_YDM) {
+    formats[count] = GNCIF_DATE_YDM;
+    menus[count].name = _("y/d/m");
+    menus[count].callback = choice_option_changed;
+    menus[count].user_data = &index;
+    count++;
+  }
+
+  g_assert(count > 1);
+  menu = gnc_build_option_menu(menus, count);
+  gtk_box_pack_start(GTK_BOX(menu_box), menu, TRUE, TRUE, 0);
+
+  gtk_widget_show_all(dialog);
+  gnome_dialog_run_and_close(GNOME_DIALOG(dialog));
+
+  return formats[index];
+}
+
+GncImportFormat
+gnc_import_choose_fmt(const char* msg, GncImportFormat fmts, gpointer data)
+		      
+{
+  GladeXML *xml;
+  GtkWidget *dialog;
+  GtkWidget *widget;
+
+  g_return_val_if_fail(fmts, FALSE);
+
+  /* if there is only one format availble, just return it */
+  if (!(fmts & (fmts-1))) {
+    return fmts;
+  }
+
+  xml = gnc_glade_xml_new("generic-import.glade", "format_picker");
+  dialog = glade_xml_get_widget(xml, "format_picker");
+  widget = glade_xml_get_widget(xml, "msg_label");
+  gtk_label_set_text(GTK_LABEL(widget), msg);
+
+  widget = glade_xml_get_widget(xml, "menu_box");
+  return add_menu_and_run_dialog(dialog, widget, fmts);
+}
--- a/src/import-export/import-parse.c
+++ b/src/import-export/import-parse.c
@ -0,0 +1,363 @@
+/*
+ * import-parse.c -- a generic "parser" API for importers..  Allows importers
+ * 	to parse dates and numbers, and provides a UI to ask for users to
+ * 	resolve ambiguities.
+ *
+ * Created by:	Derek Atkins <derek@ihtfp.com>
+ *
+ */
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <glib.h>
+#include <string.h>
+
+/* For regex */
+#include <sys/types.h>
+#include <regex.h>
+
+#include "gnc-engine-util.h"
+#include "gnc-ui-util.h"
+
+#include "import-parse.h"
+
+static short module = MOD_IMPORT;
+
+/* numeric regular expressions */
+static regex_t decimal_radix_regex;
+static regex_t comma_radix_regex;
+
+/* date regular expressions */
+static regex_t date_regex;
+static regex_t date_mdy_regex;
+static regex_t date_ymd_regex;
+
+static gboolean regex_compiled = FALSE;
+
+static void
+compile_regex(void)
+{
+  int flags = REG_EXTENDED;
+
+  /* compile the numeric regular expressions */
+  regcomp(&decimal_radix_regex,
+	  "^ *\\$?[+-]?\\$?[0-9]+ *$|^ *\\$?[+-]?\\$?[0-9]?[0-9]?[0-9]?(,[0-9][0-9][0-9])*(\\.[0-9]*)? *$|^ *\\$?[+-]?\\$?[0-9]+\\.[0-9]* *$", flags);
+  regcomp(&comma_radix_regex,
+	  "^ *\\$?[+-]?\\$?[0-9]+ *$|^ *\\$?[+-]?\\$?[0-9]?[0-9]?[0-9]?(\\.[0-9][0-9][0-9])*(,[0-9]*)? *$|^ *\\$?[+-]?\\$?[0-9]+,[0-9]* *$", flags);
+
+  /* compile the date-parsing regular expressions */
+  regcomp(&date_regex,
+	  "^ *([0-9]+) *[-/.'] *([0-9]+) *[-/.'] *([0-9]+).*$|^ *([0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9]).*$", flags);
+  regcomp(&date_mdy_regex, "([0-9][0-9])([0-9][0-9])([0-9][0-9][0-9][0-9])", flags);
+  regcomp(&date_ymd_regex, "([0-9][0-9][0-9][0-9])([0-9][0-9])([0-9][0-9])", flags);
+
+  regex_compiled = TRUE;
+}
+
+static gint
+my_strntol(const char *str, int len)
+{
+  gint res = 0;
+
+  g_return_val_if_fail(str, 0);
+  g_return_val_if_fail(len, 0);
+
+  while(len--) {
+
+    if (*str < '0' || *str > '9') {
+      str++;
+      continue;
+    }
+      
+    res *= 10;
+    res += *(str++) - '0';
+  }
+  return res;
+}
+
+/*
+ * based on a trio match (matches in spaces 1, 2, and 3), and a list
+ * of possible date formats, return the list of formats that this string
+ * could actually be.
+ */
+static GncImportFormat
+check_date_format(const char * str, regmatch_t *match, GncImportFormat fmts)
+{
+  GncImportFormat res = 0;
+  int len0=0, len1=0, len2=0;
+  int val0=0, val1=0, val2=0;
+
+  g_return_val_if_fail(match, res);
+  g_return_val_if_fail(fmts, res);
+
+  /* Compute the lengths */
+  len0 = match[1].rm_eo - match[1].rm_so;
+  len1 = match[2].rm_eo - match[2].rm_so;
+  len2 = match[3].rm_eo - match[3].rm_so;
+
+  /* compute the numeric values */
+  val0 = my_strntol(str + match[1].rm_so, len0);
+  val1 = my_strntol(str + match[2].rm_so, len1);
+  val2 = my_strntol(str + match[3].rm_so, len2);
+
+  /* Filter out the possibilities.  Hopefully only one will remain */
+
+  if (val0 > 12) import_clear_flag(fmts, GNCIF_DATE_MDY);
+  if (val0 > 31) import_clear_flag(fmts, GNCIF_DATE_DMY);
+  if (val0 < 1) {
+    import_clear_flag(fmts, GNCIF_DATE_DMY);
+    import_clear_flag(fmts, GNCIF_DATE_MDY);
+  }
+
+  if (val1 > 12) {
+    import_clear_flag(fmts, GNCIF_DATE_DMY);
+    import_clear_flag(fmts, GNCIF_DATE_YMD);
+  }
+  if (val1 > 31) {
+    import_clear_flag(fmts, GNCIF_DATE_MDY);
+    import_clear_flag(fmts, GNCIF_DATE_YDM);
+  }
+
+  if (val2 > 12) import_clear_flag(fmts, GNCIF_DATE_YDM);
+  if (val2 > 31) import_clear_flag(fmts, GNCIF_DATE_YMD);
+  if (val2 < 1) {
+    import_clear_flag(fmts, GNCIF_DATE_YMD);
+    import_clear_flag(fmts, GNCIF_DATE_YDM);
+  }
+
+  /* if we've got a 4-character year, make sure the value is greater
+   * than 1930 and less than 2100.  XXX: be sure to fix this by 2100!
+   */
+  if (len0 == 4 && (val0 < 1930 || val0 > 2100)) {
+    import_clear_flag(fmts, GNCIF_DATE_YMD);
+    import_clear_flag(fmts, GNCIF_DATE_YDM);
+  }
+  if (len2 == 4 && (val2 < 1930 || val2 > 2100)) {
+    import_clear_flag(fmts, GNCIF_DATE_MDY);
+    import_clear_flag(fmts, GNCIF_DATE_DMY);
+  }
+
+  /* If the first string has a length of only 1, then it is definitely
+   * not a year (although it could be a month or day).
+   */
+  if (len0 == 1) {
+    import_clear_flag(fmts, GNCIF_DATE_YMD);
+    import_clear_flag(fmts, GNCIF_DATE_YDM);
+  }
+
+  return fmts;
+}
+
+GncImportFormat
+gnc_import_test_numeric(const char* str, GncImportFormat fmts)
+{
+  GncImportFormat res = 0;
+
+  g_return_val_if_fail(str, fmts);
+
+  if (!regex_compiled)
+    compile_regex();
+
+  if ((fmts & GNCIF_NUM_PERIOD) && !regexec(&decimal_radix_regex, str, 0, NULL, 0))
+    res |= GNCIF_NUM_PERIOD;
+
+  if ((fmts & GNCIF_NUM_COMMA) && !regexec(&comma_radix_regex, str, 0, NULL, 0))
+    res |= GNCIF_NUM_COMMA;
+
+  return res;
+}
+
+
+GncImportFormat
+gnc_import_test_date(const char* str, GncImportFormat fmts)
+{
+  regmatch_t match[5];
+  GncImportFormat res = 0;
+
+  g_return_val_if_fail(str, fmts);
+  g_return_val_if_fail(strlen(str) > 1, fmts);
+
+  if (!regex_compiled)
+    compile_regex();
+
+  if (!regexec(&date_regex, str, 5, match, 0)) {
+    if (match[1].rm_so != -1)
+      res = check_date_format(str, match, fmts);
+    else {
+      /* Hmm, it matches XXXXXXXX, but is this YYYYxxxx or xxxxYYYY?
+       * let's try both ways and let the parser check that YYYY is
+       * valid.
+       */
+      char temp[9];
+
+      g_return_val_if_fail(match[4].rm_so != -1, fmts);
+      g_return_val_if_fail(match[4].rm_eo - match[4].rm_so == 8, fmts);
+      
+      /* make a temp copy of the XXXXXXXX string */
+      strncpy(temp, str+match[4].rm_so, 8);
+      temp[8] = '\0';
+
+      /* then check it against the ymd or mdy formats, as necessary */
+      if (((fmts & GNCIF_DATE_YDM) || (fmts & GNCIF_DATE_YMD)) &&
+	  !regexec(&date_ymd_regex, temp, 4, match, 0))
+	res |= check_date_format(temp, match, fmts);
+
+      if (((fmts & GNCIF_DATE_DMY) || (fmts & GNCIF_DATE_MDY)) &&
+	  !regexec(&date_mdy_regex, temp, 4, match, 0))
+	res |= check_date_format(temp, match, fmts);
+    }
+  }
+
+  return res;
+}
+
+gboolean
+gnc_import_parse_numeric(const char* str, GncImportFormat fmt, gnc_numeric *val)
+{
+  g_return_val_if_fail(str, FALSE);
+  g_return_val_if_fail(val, FALSE);
+  g_return_val_if_fail(fmt, FALSE);
+  g_return_val_if_fail(!(fmt & (fmt-1)), FALSE);
+
+  switch(fmt) {
+  case GNCIF_NUM_PERIOD:
+    return xaccParseAmountExtended(str, TRUE, '-', '.', ',', NULL, "$+",
+				   val, NULL);
+  case GNCIF_NUM_COMMA:
+    return xaccParseAmountExtended(str, TRUE, '-', ',', '.', NULL, "$+",
+				   val, NULL);
+  default:
+    PERR("invalid format: %d", fmt);
+    return FALSE;
+  }
+}
+
+/* Handle y2k fixes, etc.
+ * obtaining the year "00", "2000", and "19100" all mean the same thing.
+ * output is an integer representing the year in the C.E.
+ */
+static int
+fix_year(int y)
+{
+  /* two-digit numbers less than "70"  are interpretted to be post-2000. */
+  if (y < 70)
+    return (y + 2000);
+
+  /* fix a common bug in printing post-2000 dates as 19100, etc. */
+  if (y > 19000)
+    return (1900 + (y - 19000));
+
+  /* At this point we just want to make sure that this is a real date.
+   * y _should_ be a 'unix year' (which is the number of years since
+   * 1900), but it _COULD_ be a full date (1999, 2001, etc.).  At some
+   * point in the future we can't tell the difference, but are we really
+   * going to care if this code fails in 3802?
+   */
+  if (y < 1902)
+    return (y + 1900);
+
+  /* y is good as it is */
+  return y;
+}
+
+gboolean
+gnc_import_parse_date(const char *str, GncImportFormat fmt, Timespec *val)
+{
+  regmatch_t match[5];
+  char temp[9];
+  const char *datestr;
+
+  int v0 = 0, v1=0, v2=0;
+  int m=0, d=0, y=0;
+
+  g_return_val_if_fail(str, FALSE);
+  g_return_val_if_fail(val, FALSE);
+  g_return_val_if_fail(fmt, FALSE);
+  g_return_val_if_fail(!(fmt & (fmt-1)), FALSE);
+
+  if (!regexec(&date_regex, str, 5, match, 0)) {
+    if (match[1].rm_so != -1)
+      datestr = str;
+    else {
+      /* date is of the form XXXXXXX; save it to a temp string and
+       * split it based on the format, either YYYYaabb or aabbYYYY
+       */
+      g_return_val_if_fail(match[4].rm_so != -1, FALSE);
+      g_return_val_if_fail(match[4].rm_eo - match[4].rm_so == 8, FALSE);
+
+      strncpy(temp, str+match[4].rm_so, 8);
+      temp[8] = '\0';
+
+      switch (fmt) {
+      case GNCIF_DATE_DMY:
+      case GNCIF_DATE_MDY:
+	g_return_val_if_fail(!regexec(&date_mdy_regex, temp, 4, match, 0), FALSE);
+	break;
+      case GNCIF_DATE_YMD:
+      case GNCIF_DATE_YDM:
+	g_return_val_if_fail(!regexec(&date_ymd_regex, temp, 4, match, 0), FALSE);
+	break;
+      default:
+	PERR("Invalid date format provided: %d", fmt);
+	return FALSE;
+      }
+      datestr = temp;
+    }
+
+    /* datestr points to the date string, and match[123] contains the matches. */
+
+    if (match[1].rm_so == -1 || match[2].rm_so == -1 || match[3].rm_so == -1) {
+      PERR("can't interpret date %s", str);
+      return FALSE;
+    }
+
+    /* grab the numerics */
+    v0 = my_strntol(datestr + match[1].rm_so, match[1].rm_eo-match[1].rm_so);
+    v1 = my_strntol(datestr + match[2].rm_so, match[2].rm_eo-match[2].rm_so);
+    v2 = my_strntol(datestr + match[3].rm_so, match[3].rm_eo-match[3].rm_so);
+
+    switch (fmt) {
+    case GNCIF_DATE_DMY:
+      if (v0 > 0 && v0 <= 31 && v1 > 0 && v1 <= 12 && v2 > 0) {
+	d = v0; m = v1; y = v2;
+      } else
+	PERR("format is d/m/y but date is %s", str);
+      break;
+
+    case GNCIF_DATE_MDY:
+      if (v0 > 0 && v0 <= 12 && v1 > 0 && v1 <= 31 && v2 > 0) {
+	m = v0; d = v1; y = v2;
+      } else
+	PERR("format is m/d/y but date is %s", str);
+      break;
+
+    case GNCIF_DATE_YMD:
+      if (v0 > 0 && v1 > 0 && v1 <= 12 && v2 > 0 && v2 <= 31) {
+	y = v0; m = v1; d = v2;
+      } else
+	PERR("format is y/m/d but date is %s", str);
+      break;
+
+    case GNCIF_DATE_YDM:
+      if (v0 > 0 && v1 > 0 && v1 <= 31 && v2 > 0 && v2 <= 12) {
+	y = v0; d = v1; m = v2;
+      } else
+	PERR("format is y/d/m but date is %s", str);
+      break;
+
+    default:
+      PERR("invalid date format: %d", fmt);
+    }
+
+    if (!m || !d || !y)
+      return FALSE;
+
+    y = fix_year(y);
+    *val = gnc_dmy2timespec(d, m, y);
+    return TRUE;
+  }
+
+  return FALSE;
+}
--- a/src/import-export/import-parse.h
+++ b/src/import-export/import-parse.h
@ -0,0 +1,46 @@
+/*
+ * import-parse.h -- a generic "parser" API for importers..  Allows importers
+ * 	to parse dates and numbers, and provides a UI to ask for users to
+ * 	resolve ambiguities.
+ *
+ * Created by:	Derek Atkins <derek@ihtfp.com>
+ *
+ */
+
+#ifndef IMPORT_PARSE_H
+#define IMPORT_PARSE_H
+
+#include "gnc-numeric.h"
+#include "gnc-date.h"
+
+typedef enum {
+  /* number formats */
+  GNCIF_NUM_PERIOD	= (1 << 1),
+  GNCIF_NUM_COMMA	= (1 << 2),
+
+  /* date formats */
+  GNCIF_DATE_MDY	= (1 << 8),
+  GNCIF_DATE_DMY	= (1 << 9),
+  GNCIF_DATE_YMD	= (1 << 10),
+  GNCIF_DATE_YDM	= (1 << 11)
+} GncImportFormat;
+
+
+GncImportFormat gnc_import_test_numeric(const char* str, GncImportFormat fmts);
+GncImportFormat gnc_import_test_date(const char* str, GncImportFormat fmts);
+
+
+GncImportFormat gnc_import_choose_fmt(const char* msg, GncImportFormat fmts,
+				      gpointer user_data);
+
+gboolean gnc_import_parse_numeric(const char* str, GncImportFormat fmt,
+				  gnc_numeric *val);
+gboolean gnc_import_parse_date(const char *date, GncImportFormat fmt,
+			       Timespec *val);
+
+/* Set and clear flags in bit-flags */
+#define import_set_flag(i,f) (i |= f)
+#define import_clear_flag(i,f) (i &= ~f)
+
+
+#endif /* IMPORT_PARSE_H */
--- a/src/import-export/test/Makefile.am
+++ b/src/import-export/test/Makefile.am
@ -1,6 +1,45 @@
-TESTS=test-link
+AM_CFLAGS = \
+  -I${top_srcdir}/src \
+  -I${top_srcdir}/src/gnc-module \
+  -I${top_srcdir}/src/test-core \
+  -I${top_srcdir}/src/engine \
+  -I${top_srcdir}/src/app-utils \
+  -I${top_srcdir}/src/import-export \
+  ${GUILE_INCS} \
+  ${GLIB_CFLAGS} 

-noinst_PROGRAMS=test-link
+LDADD = \
+  ${top_builddir}/src/gnc-module/libgncmodule.la \
+  ${top_builddir}/src/test-core/libgncmod-test.la \
+  ../libgncmod-generic-import.la \
+  ${GLIB_LIBS}

-test_link_SOURCES=test-link.c
-test_link_LDADD=../libgncmod-generic-import.la 
+TESTS = \
+  test-link \
+  test-import-parse
+
+GNC_TEST_DEPS := @GNC_TEST_SRFI_LOAD_CMD@ \
+  --gnc-module-dir ${top_builddir}/src/gnc-module \
+  --gnc-module-dir ${top_builddir}/src/engine \
+  --gnc-module-dir ${top_builddir}/src/app-utils \
+  --gnc-module-dir ${top_builddir}/src/import-export \
+  --gnc-module-dir ${top_builddir}/src/calculation \
+  --gnc-module-dir ${top_builddir}/src/gnome-utils \
+  --gnc-module-dir ${top_srcdir}/src/gnc-module \
+  --gnc-module-dir ${top_srcdir}/src/engine \
+  --gnc-module-dir ${top_srcdir}/src/app-utils \
+  --gnc-module-dir ${top_srcdir}/src/gnome-utils \
+  --gnc-module-dir ${top_builddir}/src/gnome-utils \
+  --gnc-module-dir ${top_builddir}/src/network-utils \
+  --gnc-module-dir ${top_builddir}/src/gnome \
+  --library-dir    ${G_WRAP_LIB_DIR} \
+  --guile-load-dir ${G_WRAP_MODULE_DIR} \
+  --guile-load-dir ${top_srcdir}/src/scm \
+  --guile-load-dir ${top_srcdir}/src/import-export
+
+TESTS_ENVIRONMENT := \
+  $(shell ${top_srcdir}/src/gnc-test-env --no-exports ${GNC_TEST_DEPS})
+
+noinst_PROGRAMS = \
+  test-link \
+  test-import-parse
--- a/src/import-export/test/test-import-parse.c
+++ b/src/import-export/test/test-import-parse.c
@ -0,0 +1,163 @@
+/*
+ * test-import-parse.c -- Test the import-parse routines.
+ *
+ * Created by:	Derek Atkins <derek@ihtfp.com>
+ *
+ */
+
+#include <glib.h>
+#include <libguile.h>
+
+#include "gnc-module.h"
+#include "import-parse.h"
+
+#include "test-stuff.h"
+
+typedef struct { int y; int m; int d; } my_ymd_t;
+
+/* make sure the numbers are the same in the two sets of lists */
+const char* period_numbers[] = { " $+2000.00", "-2.00", "1,182,183.1827", NULL };
+const char* comma_numbers[] = { " $2000,00", "-2,00", "1.182.183,1827", NULL };
+const char* period_numbers_ambig[] = { "  -$1,000 ", "100.277", NULL };
+const char* comma_numbers_ambig[] = { "  -$1.000 ", "100,277", NULL };
+
+/* Make sure the strings and numbers match... */
+const char* dates_ymd[] = { "1999/12/31", "2001-6-17", "20020726",  NULL };
+my_ymd_t dates_ymd_vals[] = { {1999,12,31}, {2001,6,17}, {2002,7,26}, {0,0,0} };
+const char* dates_ydm[] = { "1999/31/12", "2001-17-6", "20012311", NULL };
+my_ymd_t dates_ydm_vals[] = { {1999,12,31}, {2001,6,17}, {2001,11,23}, {0,0,0} };
+const char* dates_mdy[] = { "1/16/2001", "12-31-1999", "01171983", NULL };
+my_ymd_t dates_mdy_vals[] = { {2001,1,16}, {1999,12,31}, {1983,1,17}, {0,0,0} };
+const char* dates_dmy[] = { "16/1/2001", "31-12-1999", "17011976", NULL };
+my_ymd_t dates_dmy_vals[] = { {2001,1,16}, {1999,12,31}, {1976,1,17}, {0,0,0} };
+
+const char* dates_yxx[] = { "99/1/6", "1999-12'10", "20010306", NULL };
+const char* dates_xxy[] = { "1/3/99", "12-10'1999", "03062001", NULL };
+
+static void
+run_check(GncImportFormat (*check_fcn)(const char*, GncImportFormat),
+	  const char *numbers[], GncImportFormat formats,
+	  const char* msg, GncImportFormat expected)
+{
+  while (*numbers) {
+    do_test(check_fcn(*numbers, formats) == expected, msg);
+    numbers++;
+  }
+}
+
+static void
+test_check_numeric(void)
+{
+  GncImportFormat fmts;
+
+  fmts = GNCIF_NUM_PERIOD | GNCIF_NUM_COMMA | GNCIF_DATE_MDY;
+  
+  run_check(gnc_import_test_numeric, period_numbers, fmts,
+	    "Period numbers", GNCIF_NUM_PERIOD);
+  run_check(gnc_import_test_numeric, comma_numbers, fmts,
+	    "Comma numbers", GNCIF_NUM_COMMA);
+
+  run_check(gnc_import_test_numeric, period_numbers_ambig, fmts,
+	    "Ambiguous Period numbers", GNCIF_NUM_PERIOD | GNCIF_NUM_COMMA);
+  run_check(gnc_import_test_numeric, comma_numbers_ambig, fmts,
+	    "Ambiguous Comma numbers", GNCIF_NUM_PERIOD | GNCIF_NUM_COMMA);
+}
+
+static void
+test_check_date(void)
+{
+  GncImportFormat fmts;
+
+  fmts = GNCIF_DATE_DMY | GNCIF_DATE_MDY | GNCIF_DATE_YMD | GNCIF_DATE_YDM;
+
+  run_check(gnc_import_test_date, dates_ymd, fmts, "y/m/d dates", GNCIF_DATE_YMD);
+  run_check(gnc_import_test_date, dates_ydm, fmts, "y/d/m dates", GNCIF_DATE_YDM);
+  run_check(gnc_import_test_date, dates_mdy, fmts, "m/d/y dates", GNCIF_DATE_MDY);
+  run_check(gnc_import_test_date, dates_dmy, fmts, "d/m/y dates", GNCIF_DATE_DMY);
+
+  run_check(gnc_import_test_date, dates_yxx, fmts, "y/x/x dates",
+	    GNCIF_DATE_YMD | GNCIF_DATE_YDM);
+  run_check(gnc_import_test_date, dates_xxy, fmts, "x/x/y dates",
+	    GNCIF_DATE_DMY | GNCIF_DATE_MDY);
+}
+
+static void
+test_numbers(const char* pstr, const char* cstr)
+{
+  gnc_numeric pval, cval;
+
+  do_test(gnc_import_parse_numeric(pstr, GNCIF_NUM_PERIOD, &pval), "Parsing Period");
+  do_test(gnc_import_parse_numeric(cstr, GNCIF_NUM_COMMA, &cval), "Parsing Comma");
+
+  do_test(gnc_numeric_equal(pval, cval), "Numbers equal?");
+}
+
+static void
+test_number_strings(const char** pstrs, const char** cstrs)
+{
+  while (*pstrs && *cstrs) {
+    test_numbers(*pstrs, *cstrs);
+    pstrs++;
+    cstrs++;
+  }
+}
+
+static void
+test_parse_numeric(void)
+{
+  test_number_strings(period_numbers, comma_numbers);
+  test_number_strings(period_numbers_ambig, comma_numbers_ambig);
+}
+
+static void
+test_date(const char* str, GncImportFormat fmt, my_ymd_t date)
+{
+  Timespec ts, ts2;;
+
+  do_test(gnc_import_parse_date(str, fmt, &ts), "Parsing date");
+  ts2 = gnc_dmy2timespec(date.d, date.m, date.y);
+  do_test(timespec_equal(&ts, &ts2), "Date Equal");
+}
+
+static void
+test_date_list(const char** strs, GncImportFormat fmt, my_ymd_t *dates)
+{
+  while (*strs && (*dates).y) {
+    test_date(*strs, fmt, *dates);
+    strs++; dates++;
+  }
+}
+
+static void
+test_parse_date(void)
+{
+  test_date_list(dates_ymd, GNCIF_DATE_YMD, dates_ymd_vals);
+  test_date_list(dates_ydm, GNCIF_DATE_YDM, dates_ydm_vals);
+  test_date_list(dates_mdy, GNCIF_DATE_MDY, dates_mdy_vals);
+  test_date_list(dates_dmy, GNCIF_DATE_DMY, dates_dmy_vals);
+}
+
+static void
+test_import_parse(void)
+{
+  test_check_numeric();
+  test_check_date();
+  test_parse_numeric();
+  test_parse_date();
+}
+
+static void
+main_helper(void *closure, int argc, char **argv)
+{
+  gnc_module_load("gnucash/import-export", 0);
+  test_import_parse();
+  print_test_results();
+  exit(get_rv());
+}
+
+int
+main(int argc, char **argv)
+{
+  scm_boot_guile(argc, argv, main_helper, NULL);
+  return 0;
+}