From f8dfd7fdda19694ec8a64b196d45866333fbdfc9 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Wed, 4 Jul 2012 18:53:58 +0530
Subject: [PATCH] Allow setting the strength of the ICU collator

---
 src/calibre/utils/icu.c  | 44 ++++++++++++++++++++++++++++++++++++---
 src/calibre/utils/icu.py | 45 +++++++++++++++++++++++++++++++++-------
 2 files changed, 79 insertions(+), 10 deletions(-)

diff --git a/src/calibre/utils/icu.c b/src/calibre/utils/icu.c
index b5805bd4e1..675d72fb87 100644
--- a/src/calibre/utils/icu.c
+++ b/src/calibre/utils/icu.c
@@ -74,6 +74,23 @@ icu_Collator_display_name(icu_Collator *self, void *closure) {
 
 // }}}
 
+// Collator.strength {{{
+static PyObject *
+icu_Collator_get_strength(icu_Collator *self, void *closure) {
+    return Py_BuildValue("i", ucol_getStrength(self->collator));
+}
+
+static int
+icu_Collator_set_strength(icu_Collator *self, PyObject *val, void *closure) {
+    if (!PyInt_Check(val)) {
+        PyErr_SetString(PyExc_TypeError, "Strength must be an integer.");
+        return -1;
+    }
+    ucol_setStrength(self->collator, (int)PyInt_AS_LONG(val));
+    return 0;
+}
+// }}}
+
 // Collator.actual_locale {{{
 static PyObject *
 icu_Collator_actual_locale(icu_Collator *self, void *closure) {
@@ -320,6 +337,12 @@ static PyGetSetDef  icu_Collator_getsetters[] = {
      (char *)"Display name of this collator in English. The name reflects the actual data source used.",
      NULL},
 
+    {(char *)"strength",
+     (getter)icu_Collator_get_strength, (setter)icu_Collator_set_strength,
+     (char *)"The strength of this collator.",
+     NULL},
+
+
     {NULL}  /* Sentinel */
 };
 
@@ -542,6 +565,7 @@ static PyMethodDef icu_methods[] = {
     {NULL}  /* Sentinel */
 };
 
+#define ADDUCONST(x) PyModule_AddIntConstant(m, #x, x)
 
 PyMODINIT_FUNC
 initicu(void) 
@@ -562,9 +586,23 @@ initicu(void)
     PyModule_AddObject(m, "Collator", (PyObject *)&icu_CollatorType);
     // uint8_t must be the same size as char
     PyModule_AddIntConstant(m, "ok", (U_SUCCESS(status) && sizeof(uint8_t) == sizeof(char)) ? 1 : 0);
-    PyModule_AddIntConstant(m, "USET_SPAN_NOT_CONTAINED", USET_SPAN_NOT_CONTAINED);
-    PyModule_AddIntConstant(m, "USET_SPAN_CONTAINED", USET_SPAN_CONTAINED);
-    PyModule_AddIntConstant(m, "USET_SPAN_SIMPLE", USET_SPAN_SIMPLE);
+
+    ADDUCONST(USET_SPAN_NOT_CONTAINED);
+    ADDUCONST(USET_SPAN_CONTAINED);
+    ADDUCONST(USET_SPAN_SIMPLE);
+    ADDUCONST(UCOL_DEFAULT);
+    ADDUCONST(UCOL_PRIMARY);
+    ADDUCONST(UCOL_SECONDARY);
+    ADDUCONST(UCOL_TERTIARY);
+    ADDUCONST(UCOL_DEFAULT_STRENGTH);
+    ADDUCONST(UCOL_QUATERNARY);
+    ADDUCONST(UCOL_IDENTICAL);
+    ADDUCONST(UCOL_OFF);
+    ADDUCONST(UCOL_ON);
+    ADDUCONST(UCOL_SHIFTED);
+    ADDUCONST(UCOL_NON_IGNORABLE);
+    ADDUCONST(UCOL_LOWER_FIRST);
+    ADDUCONST(UCOL_UPPER_FIRST);
 
 }
 // }}}
diff --git a/src/calibre/utils/icu.py b/src/calibre/utils/icu.py
index ead820f066..b5e17042ae 100644
--- a/src/calibre/utils/icu.py
+++ b/src/calibre/utils/icu.py
@@ -65,11 +65,18 @@ def py_find(pattern, source):
         return pos, len(pattern)
     return -1, -1
 
-def icu_find(collator, pattern, source):
+def icu_find(collator, pattern, source, strength=None):
+    if strength is not None:
+        ostrength = collator.strength
+        collator.strength = strength
     try:
-        return collator.find(pattern, source)
-    except TypeError:
-        return collator.find(unicode(pattern), unicode(source))
+        try:
+            return collator.find(pattern, source)
+        except TypeError:
+            return collator.find(unicode(pattern), unicode(source))
+    finally:
+        if strength is not None:
+            collator.strength = ostrength
 
 def py_case_sensitive_sort_key(obj):
     if not obj:
@@ -81,10 +88,20 @@ def icu_case_sensitive_sort_key(collator, obj):
         return _none2
     return collator.sort_key(obj)
 
-def icu_strcmp(collator, a, b):
-    return collator.strcmp(lower(a), lower(b))
+def icu_strcmp(collator, a, b, strength=None):
+    if strength is not None:
+        ostrength = collator.strength
+        collator.strength = strength
+    try:
+        s = collator.strength
+        if s >= _icu.UCOL_TERTIARY:
+            a, b = lower(a), lower(b)
+        return collator.strcmp(a, b)
+    finally:
+        if strength is not None:
+            collator.strength = ostrength
 
-def py_strcmp(a, b):
+def py_strcmp(a, b, strength=None):
     return cmp(a.lower(), b.lower())
 
 def icu_case_sensitive_strcmp(collator, a, b):
@@ -161,6 +178,20 @@ contractions = ((lambda : {}) if _icu_not_ok else (partial(icu_contractions,
 span_contractions = (py_span_contractions if _icu_not_ok else
         icu_span_contractions)
 
+def primary_strcmp(a, b):
+    'strcmp that ignores case and accents on letters'
+    if _icu_not_ok:
+        from calibre.utils.filenames import ascii_text
+        return py_strcmp(ascii_text(a), ascii_text(b))
+    return icu_strcmp(_collator, a, b, _icu.UCOL_PRIMARY)
+
+def primary_find(pat, src):
+    'find that ignores case and accents on letters'
+    if _icu_not_ok:
+        from calibre.utils.filenames import ascii_text
+        return py_find(ascii_text(pat), ascii_text(src))
+    return icu_find(_collator, pat, src, _icu.UCOL_PRIMARY)
+
 ################################################################################
 
 def test(): # {{{