From 589c41003f6ebf8daa97251773251e6d4655d037 Mon Sep 17 00:00:00 2001
From: wallace <11wallace11@gmail.com>
Date: Tue, 24 Aug 2021 01:19:42 +0300
Subject: [PATCH 1/2] Add hebrew document language
---
docker/dockerfiles/joex.dockerfile | 1 +
.../scala/docspell/analysis/date/DateFind.scala | 1 +
.../docspell/analysis/date/MonthName.scala | 17 +++++++++++++++++
.../main/scala/docspell/common/Language.scala | 9 ++++++++-
.../main/scala/docspell/ftssolr/SolrSetup.scala | 8 +++++++-
modules/webapp/src/main/elm/Data/Language.elm | 8 ++++++++
.../src/main/elm/Messages/Data/Language.elm | 6 ++++++
7 files changed, 48 insertions(+), 2 deletions(-)
diff --git a/docker/dockerfiles/joex.dockerfile b/docker/dockerfiles/joex.dockerfile
index 0d105e98..05a49367 100644
--- a/docker/dockerfiles/joex.dockerfile
+++ b/docker/dockerfiles/joex.dockerfile
@@ -30,6 +30,7 @@ RUN JDKPKG="openjdk11"; \
tesseract-ocr-data-ron \
tesseract-ocr-data-lav \
tesseract-ocr-data-jpn \
+ tesseract-ocr-data-heb \
unpaper \
wkhtmltopdf \
libreoffice \
diff --git a/modules/analysis/src/main/scala/docspell/analysis/date/DateFind.scala b/modules/analysis/src/main/scala/docspell/analysis/date/DateFind.scala
index 1257a5c9..d681eefa 100644
--- a/modules/analysis/src/main/scala/docspell/analysis/date/DateFind.scala
+++ b/modules/analysis/src/main/scala/docspell/analysis/date/DateFind.scala
@@ -102,6 +102,7 @@ object DateFind {
case Language.Dutch => dmy.or(ymd).or(mdy)
case Language.Latvian => dmy.or(lavLong).or(ymd)
case Language.Japanese => ymd
+ case Language.Hebrew => dmy
}
p.read(parts) match {
case Result.Success(sds, _) =>
diff --git a/modules/analysis/src/main/scala/docspell/analysis/date/MonthName.scala b/modules/analysis/src/main/scala/docspell/analysis/date/MonthName.scala
index 63d51a9a..5f13f302 100644
--- a/modules/analysis/src/main/scala/docspell/analysis/date/MonthName.scala
+++ b/modules/analysis/src/main/scala/docspell/analysis/date/MonthName.scala
@@ -52,6 +52,8 @@ object MonthName {
latvian
case Language.Japanese =>
japanese
+ case Language.Hebrew =>
+ hebrew
}
private val numbers = List(
@@ -307,4 +309,19 @@ object MonthName {
List("11", "十一"),
List("12", "十二")
)
+
+ private val hebrew = List(
+ List("1", "ינואר"),
+ List("2", "פברואר"),
+ List("3", "מרץ"),
+ List("4", "אפריל"),
+ List("5", "מאי"),
+ List("6", "יוני"),
+ List("7", "יולי"),
+ List("8", "אוגוסט"),
+ List("9", "ספטמבר"),
+ List("10", "אוקטובר"),
+ List("11", "נובמבר"),
+ List("12", "דצמבר")
+ )
}
diff --git a/modules/common/src/main/scala/docspell/common/Language.scala b/modules/common/src/main/scala/docspell/common/Language.scala
index d46aba3a..db233296 100644
--- a/modules/common/src/main/scala/docspell/common/Language.scala
+++ b/modules/common/src/main/scala/docspell/common/Language.scala
@@ -113,6 +113,12 @@ object Language {
val iso3 = "jpn"
}
+ case object Hebrew extends Language {
+ val iso2 = "he"
+ val iso3 = "heb"
+ }
+
+
val all: List[Language] =
List(
German,
@@ -130,7 +136,8 @@ object Language {
Russian,
Romanian,
Latvian,
- Japanese
+ Japanese,
+ Hebrew
)
def fromString(str: String): Either[String, Language] = {
diff --git a/modules/fts-solr/src/main/scala/docspell/ftssolr/SolrSetup.scala b/modules/fts-solr/src/main/scala/docspell/ftssolr/SolrSetup.scala
index 6aa7dd36..8ebd323b 100644
--- a/modules/fts-solr/src/main/scala/docspell/ftssolr/SolrSetup.scala
+++ b/modules/fts-solr/src/main/scala/docspell/ftssolr/SolrSetup.scala
@@ -114,7 +114,13 @@ object SolrSetup {
"Add japanese content field",
addContentField(Language.Japanese)
),
- SolrMigration.reIndexAll(15, "Re-Index after adding japanese content field")
+ SolrMigration.reIndexAll(15, "Re-Index after adding japanese content field"),
+ SolrMigration[F](
+ 16,
+ "Add hebrew content field",
+ addContentField(Language.Hebrew)
+ ),
+ SolrMigration.reIndexAll(17, "Re-Index after adding hebrew content field")
)
def addFolderField: F[Unit] =
diff --git a/modules/webapp/src/main/elm/Data/Language.elm b/modules/webapp/src/main/elm/Data/Language.elm
index 41522878..3b7970bc 100644
--- a/modules/webapp/src/main/elm/Data/Language.elm
+++ b/modules/webapp/src/main/elm/Data/Language.elm
@@ -30,6 +30,7 @@ type Language
| Dutch
| Latvian
| Japanese
+ | Hebrew
fromString : String -> Maybe Language
@@ -82,6 +83,9 @@ fromString str =
else if str == "jpn" || str == "ja" || str == "japanese" then
Just Japanese
+ else if str == "heb" || str == "he" || str == "hebrew" then
+ Just Hebrew
+
else
Nothing
@@ -137,6 +141,9 @@ toIso3 lang =
Japanese ->
"jpn"
+ Hebrew ->
+ "heb"
+
all : List Language
all =
@@ -156,4 +163,5 @@ all =
, Romanian
, Latvian
, Japanese
+ , Hebrew
]
diff --git a/modules/webapp/src/main/elm/Messages/Data/Language.elm b/modules/webapp/src/main/elm/Messages/Data/Language.elm
index 60070c56..6e7fac1e 100644
--- a/modules/webapp/src/main/elm/Messages/Data/Language.elm
+++ b/modules/webapp/src/main/elm/Messages/Data/Language.elm
@@ -64,6 +64,9 @@ gb lang =
Japanese ->
"Japanese"
+ Hebrew ->
+ "Hebrew"
+
de : Language -> String
de lang =
@@ -115,3 +118,6 @@ de lang =
Japanese ->
"Japanisch"
+
+ Hebrew ->
+ "Hebräisch"
From eab7fe57d278c508497496e602a6c0305479667a Mon Sep 17 00:00:00 2001
From: wallace <11wallace11@gmail.com>
Date: Tue, 24 Aug 2021 02:24:20 +0300
Subject: [PATCH 2/2] Fix formatting with scalafmt
---
modules/common/src/main/scala/docspell/common/Language.scala | 1 -
1 file changed, 1 deletion(-)
diff --git a/modules/common/src/main/scala/docspell/common/Language.scala b/modules/common/src/main/scala/docspell/common/Language.scala
index db233296..3a6c9eb0 100644
--- a/modules/common/src/main/scala/docspell/common/Language.scala
+++ b/modules/common/src/main/scala/docspell/common/Language.scala
@@ -118,7 +118,6 @@ object Language {
val iso3 = "heb"
}
-
val all: List[Language] =
List(
German,