[pango/log-attr-tweaks: 33/35] Add api to validate log attrs
- From: Matthias Clasen <matthiasc src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [pango/log-attr-tweaks: 33/35] Add api to validate log attrs
- Date: Thu, 26 Aug 2021 12:22:26 +0000 (UTC)
commit 801e93cfa39a450d28a35ad8a36f84faab19fd36
Author: Matthias Clasen <mclasen redhat com>
Date: Sat Aug 21 15:41:39 2021 -0400
Add api to validate log attrs
This is based on code that previously lived in
tests/validate-test-boundaries.c
Fixes: #129
pango/break.c | 492 +++++++++++++++++++++++++++++++++++++++++-
pango/pango-break.h | 22 ++
tests/meson.build | 2 +-
tests/test-break.c | 1 -
tests/validate-log-attrs.c | 528 ---------------------------------------------
tests/validate-log-attrs.h | 52 -----
6 files changed, 514 insertions(+), 583 deletions(-)
---
diff --git a/pango/break.c b/pango/break.c
index 0cb6666d..20756506 100644
--- a/pango/break.c
+++ b/pango/break.c
@@ -1759,7 +1759,9 @@ remove_breaks_from_range (const char *text,
ch == 0x007C;
/* ZWS sequence */
- if (after_zws && bt != G_UNICODE_BREAK_SPACE)
+ if (after_zws &&
+ bt != G_UNICODE_BREAK_SPACE &&
+ bt != G_UNICODE_BREAK_ZERO_WIDTH_SPACE)
log_attrs[pos].is_line_break = TRUE;
after_zws = bt == G_UNICODE_BREAK_ZERO_WIDTH_SPACE ||
@@ -2141,6 +2143,428 @@ tailor_break (const char *text,
return res;
}
+/* }}} */
+/* {{{ Validation */
+
+G_DEFINE_QUARK(pango-validate-error-quark, pango_validate_error)
+
+typedef gboolean (* CharForeachFunc) (int pos,
+ gunichar wc,
+ gunichar prev_wc,
+ gunichar next_wc,
+ GUnicodeType type,
+ GUnicodeType prev_type,
+ GUnicodeType next_type,
+ const PangoLogAttr *attr,
+ const PangoLogAttr *prev_attr,
+ const PangoLogAttr *next_attr,
+ gboolean *after_zws,
+ GError **error);
+
+static gboolean
+log_attr_foreach (const char *text,
+ int length,
+ const PangoLogAttr *attrs,
+ int attrs_len,
+ CharForeachFunc func,
+ GError **error)
+{
+ const gchar *next = text;
+ const gchar *end = text + length;
+ gint i = 0;
+ gunichar prev_wc;
+ gunichar next_wc;
+ GUnicodeType prev_type;
+ GUnicodeType next_type;
+ gboolean after_zws;
+
+ if (next == end)
+ goto done;
+
+ prev_type = (GUnicodeType) -1;
+ prev_wc = 0;
+
+ next_wc = g_utf8_get_char (next);
+ next_type = g_unichar_type (next_wc);
+
+ after_zws = FALSE;
+
+ while (next_wc != 0)
+ {
+ GUnicodeType type;
+ gunichar wc;
+
+ wc = next_wc;
+ type = next_type;
+
+ next = g_utf8_next_char (next);
+
+ if (next >= end)
+ next_wc = 0;
+ else
+ next_wc = g_utf8_get_char (next);
+
+ if (next_wc)
+ next_type = g_unichar_type (next_wc);
+
+ if (!func (i,
+ wc, prev_wc, next_wc,
+ type, prev_type, next_type,
+ &attrs[i],
+ i != 0 ? &attrs[i - 1] : NULL,
+ &attrs[i + 1],
+ &after_zws,
+ error))
+ return FALSE;
+
+ prev_type = type;
+ prev_wc = wc;
+ i++;
+ }
+
+done:
+ return TRUE;
+}
+
+static gboolean
+check_line_char (int pos,
+ gunichar wc,
+ gunichar prev_wc,
+ gunichar next_wc,
+ GUnicodeType type,
+ GUnicodeType prev_type,
+ GUnicodeType next_type,
+ const PangoLogAttr *attr,
+ const PangoLogAttr *prev_attr,
+ const PangoLogAttr *next_attr,
+ gboolean *after_zws,
+ GError **error)
+{
+ GUnicodeBreakType break_type;
+ GUnicodeBreakType prev_break_type;
+
+ break_type = g_unichar_break_type (wc);
+
+ if (prev_wc)
+ prev_break_type = g_unichar_break_type (prev_wc);
+ else
+ prev_break_type = G_UNICODE_BREAK_UNKNOWN;
+
+ if (prev_break_type == G_UNICODE_BREAK_ZERO_WIDTH_SPACE ||
+ (prev_break_type == G_UNICODE_BREAK_SPACE && *after_zws))
+ *after_zws = TRUE;
+ else
+ *after_zws = FALSE;
+
+ if (wc == '\n' && prev_wc == '\r')
+ {
+ if (attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Do not break between \\r and \\n (LB5)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (prev_wc == 0 && wc != 0)
+ {
+ if (attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Do not break before first char (LB2)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (next_wc == 0)
+ {
+ if (!next_attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Always break after the last char (LB3)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (prev_break_type == G_UNICODE_BREAK_MANDATORY)
+ {
+ if (!attr->is_mandatory_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Always break after hard line breaks (LB4)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (prev_break_type == G_UNICODE_BREAK_CARRIAGE_RETURN ||
+ prev_break_type == G_UNICODE_BREAK_LINE_FEED ||
+ prev_break_type == G_UNICODE_BREAK_NEXT_LINE)
+ {
+ if (!attr->is_mandatory_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Always break after CR, LF and NL (LB5)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (break_type == G_UNICODE_BREAK_MANDATORY ||
+ break_type == G_UNICODE_BREAK_CARRIAGE_RETURN ||
+ break_type == G_UNICODE_BREAK_LINE_FEED ||
+ break_type == G_UNICODE_BREAK_NEXT_LINE)
+ {
+ if (attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Do not break before hard line beaks (LB6)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (break_type == G_UNICODE_BREAK_SPACE ||
+ break_type == G_UNICODE_BREAK_ZERO_WIDTH_SPACE)
+ {
+ if (attr->is_line_break && prev_attr != NULL &&
+ !attr->is_mandatory_break &&
+ !(next_wc && g_unichar_break_type (next_wc) == G_UNICODE_BREAK_COMBINING_MARK))
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Can't break before a space unless mandatory precedes or combining mark
follows (LB7)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (break_type != G_UNICODE_BREAK_ZERO_WIDTH_SPACE &&
+ break_type != G_UNICODE_BREAK_SPACE &&
+ *after_zws)
+ {
+ if (!attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Break before a char following ZWS, even if spaces intervene (LB8)", wc,
pos);
+ return FALSE;
+ }
+ }
+
+ if (break_type == G_UNICODE_BREAK_ZERO_WIDTH_JOINER)
+ {
+ if (attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Do not break after ZWJ (LB8a)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ /* TODO: check LB9 */
+
+ if (prev_break_type == G_UNICODE_BREAK_WORD_JOINER ||
+ break_type == G_UNICODE_BREAK_WORD_JOINER)
+ {
+ if (attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Do not break before or after WJ (LB11)", wc, pos);
+ return FALSE;
+ }
+ }
+
+ if (prev_break_type == G_UNICODE_BREAK_NON_BREAKING_GLUE)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Do not break after GL (LB12)", wc, pos);
+ return FALSE;
+ }
+
+ /* internal consistency */
+
+ if (attr->is_mandatory_break && !attr->is_line_break)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_BREAK,
+ "char %#x %d: Mandatory breaks must also be marked as regular breaks", wc, pos);
+ return FALSE;
+ }
+
+ return TRUE;
+}
+
+static gboolean
+check_line_invariants (const char *text,
+ int length,
+ const PangoLogAttr *attrs,
+ int attrs_len,
+ GError **error)
+{
+ return log_attr_foreach (text, length,
+ attrs, attrs_len,
+ check_line_char, error);
+}
+
+static gboolean
+check_grapheme_invariants (const char *text,
+ int length,
+ const PangoLogAttr *attrs,
+ int attrs_len,
+ GError **error)
+{
+ return TRUE;
+}
+
+static gboolean
+check_word_invariants (const char *text,
+ int length,
+ const PangoLogAttr *attrs,
+ int attrs_len,
+ GError **error)
+{
+ enum {
+ AFTER_START,
+ AFTER_END
+ } state = AFTER_END;
+
+ for (int i = 0; i < attrs_len; i++)
+ {
+ /* Check that word starts and ends are alternating */
+ switch (state)
+ {
+ case AFTER_END:
+ if (attrs[i].is_word_start)
+ {
+ if (attrs[i].is_word_end)
+ state = AFTER_END;
+ else
+ state = AFTER_START;
+ break;
+ }
+ if (attrs[i].is_word_end)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_WORD,
+ "char %d: Unexpected word end", i);
+ return FALSE;
+ }
+ break;
+
+ case AFTER_START:
+ if (attrs[i].is_word_end)
+ {
+ if (attrs[i].is_word_start)
+ state = AFTER_START;
+ else
+ state = AFTER_END;
+ break;
+ }
+ if (attrs[i].is_word_start)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_WORD,
+ "char %d: Unexpected word start", i);
+ return FALSE;
+ }
+ break;
+
+ default:
+ g_assert_not_reached ();
+ }
+ }
+
+ return TRUE;
+}
+
+static gboolean
+check_sentence_invariants (const char *text,
+ int length,
+ const PangoLogAttr *attrs,
+ int attrs_len,
+ GError **error)
+{
+ enum {
+ AFTER_START,
+ AFTER_END
+ } state = AFTER_END;
+
+ for (int i = 0; i < attrs_len; i++)
+ {
+ /* Check that word starts and ends are alternating */
+ switch (state)
+ {
+ case AFTER_END:
+ if (attrs[i].is_sentence_start)
+ {
+ if (attrs[i].is_sentence_end)
+ state = AFTER_END;
+ else
+ state = AFTER_START;
+ break;
+ }
+ if (attrs[i].is_sentence_end)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_SENTENCE,
+ "char %d: Unexpected sentence end", i);
+ return FALSE;
+ }
+ break;
+
+ case AFTER_START:
+ if (attrs[i].is_sentence_end)
+ {
+ if (attrs[i].is_sentence_start)
+ state = AFTER_START;
+ else
+ state = AFTER_END;
+ break;
+ }
+ if (attrs[i].is_sentence_start)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_SENTENCE,
+ "char %d: Unexpected sentence start", i);
+ return FALSE;
+ }
+ break;
+
+ default:
+ g_assert_not_reached ();
+ }
+ }
+
+ return TRUE;
+}
+
+static gboolean
+check_space_invariants (const char *text,
+ int length,
+ const PangoLogAttr *log_attrs,
+ int attrs_len,
+ GError **error)
+{
+ for (int i = 0; i < attrs_len; i++)
+ {
+ if (log_attrs[i].is_expandable_space && !log_attrs[i].is_white)
+ {
+ g_set_error (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_SPACE,
+ "char %d: Expandable space must be space", i);
+ return FALSE;
+ }
+ }
+
+ return TRUE;
+}
+
/* }}} */
/* {{{ Public API */
@@ -2377,6 +2801,72 @@ pango_get_log_attrs (const char *text,
attrs_len);
}
+/**
+ * pango_validate_log_attrs:
+ * @text: text to which @log_attrs belong
+ * @length: length of @text
+ * @log_attrs: `PangoLogAttr` array to validate
+ * @attrs_len: length of @log_attrs
+ *
+ * Apply sanity checks to @log_attrs.
+ *
+ * This function checks some conditions that Pango
+ * relies on. It is not guaranteed to be an exhaustive
+ * validity test. Currentlty, it checks that
+ *
+ * - There's no break before the first char
+ * - Mandatory breaks are line breaks
+ * - Line breaks are char breaks
+ * - Lines aren't broken between \\r and \\n
+ * - Lines aren't broken before a space (unless the break
+ * is mandatory, or the space precedes a combining mark)
+ * - Lines aren't broken between two open punctuation
+ * or between two close punctuation characters
+ * - Lines aren't broken between a letter and a quotation mark
+ * - Word starts and ends alternate
+ * - Sentence starts and ends alternate
+ * - Expandable spaces are spaces
+ *
+ * Returns: %TRUE if @log_attrs are valid
+ *
+ * Since: 1.50
+ */
+gboolean
+pango_validate_log_attrs (const char *text,
+ int length,
+ const PangoLogAttr *log_attrs,
+ int attrs_len,
+ GError **error)
+{
+ int n_chars;
+
+ n_chars = g_utf8_strlen (text, length);
+ if (attrs_len != n_chars + 1)
+ {
+ g_set_error_literal (error,
+ PANGO_VALIDATE_ERROR, PANGO_VALIDATE_ERROR_FAILED,
+ "Array has wrong length");
+ return FALSE;
+ }
+
+ if (!check_line_invariants (text, length, log_attrs, attrs_len, error))
+ return FALSE;
+
+ if (!check_grapheme_invariants (text, length, log_attrs, attrs_len, error))
+ return FALSE;
+
+ if (!check_word_invariants (text, length, log_attrs, attrs_len, error))
+ return FALSE;
+
+ if (!check_sentence_invariants (text, length, log_attrs, attrs_len, error))
+ return FALSE;
+
+ if (!check_space_invariants (text, length, log_attrs, attrs_len, error))
+ return FALSE;
+
+ return TRUE;
+}
+
/* }}} */
/* vim:set foldmethod=marker expandtab: */
diff --git a/pango/pango-break.h b/pango/pango-break.h
index 5d791e27..b29d713a 100644
--- a/pango/pango-break.h
+++ b/pango/pango-break.h
@@ -137,6 +137,28 @@ void pango_attr_break (const char *text,
PangoLogAttr *attrs,
int attrs_len);
+#define PANGO_VALIDATE_ERROR (pango_validate_error_quark ())
+
+typedef enum
+{
+ PANGO_VALIDATE_ERROR_FAILED,
+ PANGO_VALIDATE_ERROR_BREAK,
+ PANGO_VALIDATE_ERROR_GRAPHEME,
+ PANGO_VALIDATE_ERROR_WORD,
+ PANGO_VALIDATE_ERROR_SENTENCE,
+ PANGO_VALIDATE_ERROR_SPACE
+} PangoValidateError;
+
+PANGO_AVAILABLE_IN_1_50
+GQuark pango_validate_error_quark (void);
+
+PANGO_AVAILABLE_IN_1_50
+gboolean pango_validate_log_attrs (const char *text,
+ int length,
+ const PangoLogAttr *log_attrs,
+ int attrs_len,
+ GError **error);
+
G_END_DECLS
#endif /* __PANGO_BREAK_H__ */
diff --git a/tests/meson.build b/tests/meson.build
index 90bb94ba..4776b2c9 100644
--- a/tests/meson.build
+++ b/tests/meson.build
@@ -52,7 +52,7 @@ if cairo_dep.found()
[ 'testmisc', [ 'testmisc.c' ], [ libpangocairo_dep, libpangoft2_dep, glib_dep, harfbuzz_dep ] ],
[ 'cxx-test', [ 'cxx-test.cpp' ], [ libpangocairo_dep, gobject_dep, harfbuzz_dep ] ],
[ 'test-harfbuzz', [ 'test-harfbuzz.c' ], [ libpangocairo_dep, gobject_dep, harfbuzz_dep ] ],
- [ 'test-break', [ 'test-break.c', 'test-common.c', 'validate-log-attrs.c' ], [libpangocairo_dep,
glib_dep, harfbuzz_dep ] ]
+ [ 'test-break', [ 'test-break.c', 'test-common.c' ], [libpangocairo_dep, glib_dep, harfbuzz_dep ] ]
]
if pango_cairo_backends.contains('png')
diff --git a/tests/test-break.c b/tests/test-break.c
index fecca168..672d44e1 100644
--- a/tests/test-break.c
+++ b/tests/test-break.c
@@ -30,7 +30,6 @@
#include "config.h"
#include <pango/pangocairo.h>
#include "test-common.h"
-#include "validate-log-attrs.h"
static PangoContext *context;
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]
