xapian: update stopwords from Xapian upstream repo

imap/xapian_wrap.cpp

@@ -57,7 +57,7 @@ static Xapian::Stopper *get_stopper()
         struct buf buf = BUF_INITIALIZER;
         buf_setcstr(&buf, swpath);
         // XXX doesn't play nice with WIN32 paths
-        buf_appendcstr(&buf, "/english.list");
+        buf_appendcstr(&buf, "/english.txt");
 
         // Open the stopword file
         errno = 0;

languages/stopwords/arabic.list → languages/stopwords/arabic.txt

@@ -1,3 +1,9 @@
+
+ | An Arabic stop word list. Comments begin with vertical bar. Each stop
+ |  word is at the start of a line.
+ |  (This is not an official Snowball stop word list, basically inspired from
+ |   Arabic Stop Words Project)
+
 إذ
 إذا
 إذما
@@ -25,18 +31,20 @@
 إليكن
 أم
 أما
+أما
 إما
 أن
 إن
-أنا
 إنا
+أنا
 أنت
 أنتم
 أنتما
 أنتن
 إنما
 إنه
 أنى
+أنى
 آه
 آها
 أو
@@ -45,17 +53,19 @@
 أوه
 آي
 أي
+أيها
 إي
 أين
+أين
 أينما
 إيه
-أيها
 بخ
 بس
 بعد
 بعض
 بك
 بكم
+بكم
 بكما
 بكن
 بل
@@ -70,8 +80,8 @@
 بهما
 بهن
 بي
-بيد
 بين
+بيد                               | though
 تلك
 تلكم
 تلكما
@@ -108,7 +118,7 @@
 ذينك
 ريث
 سوف
-سوى
+سوى                        | except
 شتان
 عدا
 عسى
@@ -119,7 +129,7 @@
 عما
 عن
 عند
-غير
+غير                              | except
 فإذا
 فإن
 فلا
@@ -144,13 +154,14 @@
 كليكما
 كليهما
 كم
+كم
 كما
 كي
 كيت
 كيف
 كيفما
 لا
-لاسيما
+لاسيما                           | especially
 لدى
 لست
 لستم
@@ -188,15 +199,15 @@
 ليسوا
 ما
 ماذا
-متى
+متى                              | when
 مذ
 مع
 مما
 ممن
 من
-منذ
 منه
 منها
+منذ
 مه
 مهما
 نحن
@@ -229,10 +240,10 @@
 هيا
 هيت
 هيهات
-وإذ
-وإذا
 والذي
 والذين
+وإذ
+وإذا
 وإن
 ولا
 ولكن

languages/stopwords/danish.txt

@@ -0,0 +1,102 @@
+
+ | A Danish stop word list. Comments begin with vertical bar. Each stop
+ | word is at the start of a line.
+
+ | This is a ranked list (commonest to rarest) of stopwords derived from
+ | a large text sample.
+
+
+og           | and
+i            | in
+jeg          | I
+det          | that (dem. pronoun)/it (pers. pronoun)
+at           | that (in front of a sentence)/to (with infinitive)
+en           | a/an
+den          | it (pers. pronoun)/that (dem. pronoun)
+til          | to/at/for/until/against/by/of/into, more
+er           | present tense of "to be"
+som          | who, as
+på           | on/upon/in/on/at/to/after/of/with/for, on
+de           | they
+med          | with/by/in, along
+han          | he
+af           | of/by/from/off/for/in/with/on, off
+for          | at/for/to/from/by/of/ago, in front/before, because
+ikke         | not
+der          | who/which, there/those
+var          | past tense of "to be"
+mig          | me/myself
+sig          | oneself/himself/herself/itself/themselves
+men          | but
+et           | a/an/one, one (number), someone/somebody/one
+har          | present tense of "to have"
+om           | round/about/for/in/a, about/around/down, if
+vi           | we
+min          | my
+havde        | past tense of "to have"
+ham          | him
+hun          | she
+nu           | now
+over         | over/above/across/by/beyond/past/on/about, over/past
+da           | then, when/as/since
+fra          | from/off/since, off, since
+du           | you
+ud           | out
+sin          | his/her/its/one's
+dem          | them
+os           | us/ourselves
+op           | up
+man          | you/one
+hans         | his
+hvor         | where
+eller        | or
+hvad         | what
+skal         | must/shall etc.
+selv         | myself/yourself/herself/ourselves etc., even
+her          | here
+alle         | all/everyone/everybody etc.
+vil          | will (verb)
+blev         | past tense of "to stay/to remain/to get/to become"
+kunne        | could
+ind          | in
+når          | when
+være         | present tense of "to be"
+dog          | however/yet/after all
+noget        | something
+ville        | would
+jo           | you know/you see (adv), yes
+deres        | their/theirs
+efter        | after/behind/according to/for/by/from, later/afterwards
+ned          | down
+skulle       | should
+denne        | this
+end          | than
+dette        | this
+mit          | my/mine
+også         | also
+under        | under/beneath/below/during, below/underneath
+have         | have
+dig          | you
+anden        | other
+hende        | her
+mine         | my
+alt          | everything
+meget        | much/very, plenty of
+sit          | his, her, its, one's
+sine         | his, her, its, one's
+vor          | our
+mod          | against
+disse        | these
+hvis         | if
+din          | your/yours
+nogle        | some
+hos          | by/at
+blive        | be/become
+mange        | many
+ad           | by/through
+bliver       | present tense of "to be/to become"
+hendes       | her/hers
+været        | be
+thi          | for (conj)
+jer          | you
+sådan        | such, like this/like that