Support batch mode exports. Reverted bad commit.

Author: Ole Liabø

formats/epjark.conf

@@ -48,10 +48,16 @@ INFOVIEW_FOLDER_TYPE_REGEXP_NN=rapportMappe
 ;   1. Node regexp
 ;   2. Pages wildcard
 ;   3. OCR wildcard
-INFO_VIEW_JOURNAL_TYPE_REGEXP=pasientjournal@%rootdir%/../objekter/????????_%fanearkidentifikator%_?_?.jpg@%rootdir%/../derivater/????????_%fanearkidentifikator%_?_?.xml
+INFO_VIEW_JOURNAL_TYPE_REGEXP=pasientjournal@%rootdir%/../objekter/????????_%fanearkidentifikator%_*_?.jpg@%rootdir%/../derivater/????????_%fanearkidentifikator%_*_?.xml
 
 ; Tool responsible for converting journal pages and optional ocr files into PDF document
 ; %FILESFILE% - File containing pages to be included, one page per line: <page-image-file>;<page-ocr-file>
 ; %OUT% - Create PDF filename
 ; %TEMP% - Output folder 
 JOURNAL_PDF_CREATOR_TOOL="./pdf/create-pdf.cmd %OUT% %FILESFILE% %TEMP%"
+
+; Auto select nodes will be selected when importing XML
+TREEVIEW_AUTO_SELECT_REGEXP=pasientjournal@diagnose
+
+; Auto collapse nodes will be collapsed when importing XML
+TREEVIEW_AUTO_COLLAPSE_REGEXP=derivater@objekter@metadata

insight.conf

@@ -39,11 +39,12 @@ REPORTS_DIR=".\\rapporter"
 
 ; Dataset to load at startup, useful for testing
 ;STARTUP_LOAD_FILE=../arkivstruktur2.xml
-;STARTUP_LOAD_FILE=arkivstruktur2.xml
+;STARTUP_LOAD_FILE="./testdata/noark-5/arkivstruktur.xml"
 ;STARTUP_LOAD_FILE=arkivstruktur.xml" 
 ;STARTUP_LOAD_FILE=e:/archivator/innsyn/noark5-stor/arkivstruktur.xml" 
 ;STARTUP_LOAD_FILE=e:/archivator/innsyn/noark5-stor/arkivstruktur-medium.xml" 
-
+; Format of startup file
+;STARTUP_LOAD_FILE_FORMAT=
 
 
 
@@ -64,5 +65,8 @@ INDEXER_TOOL=index.cmd %REPORTS_DIR% %ATTACHMENT_DIR% %NAME%
 ;SEARCH_DEAMON_TOOL=searchd --config %CONFIG_FILE% --nodetach --logdebugvv
 SEARCH_DEAMON_TOOL=searchd --config %CONFIG_FILE% --logdebugvv
 
-; Supress question box "attachments should be parsed? Y/N"
-;ATTACHMENT_PARSING_DONT_ASK_USER=true
+; Parse attachments:
+;  - YES
+;  - NO 
+;  - ASK: Display question box "attachments should be parsed? Y/N"
+;ATTACHMENT_PARSING=ASK

insight.pro

@@ -185,6 +185,7 @@ include( src/thirdparty/yxml/yxml.pri )
 
 SOURCES          +=   src/main.cpp \
                       src/drunguard.cpp \
+                      src/dregexp.cpp \
                       src/dxmlparser.cpp \
                       src/ddirparser.cpp \
                       src/dattachmentindexer.cpp \

lesmeg.org

@@ -185,6 +185,13 @@ Ved feilrapportering legg ved loggfiler og ved programkrasj /insight.dmp/ hvis d
 
 * Historikk
 
+** innsyn-v1.2.0-beta3
+*** Nye funksjoner
+ - Støtte for automatisk nodevalg, settes per format med /TREEVIEW_AUTO_SELECT_REGEXP/ variablen.
+ - Støtte for automatisk skjuling av barne noder ved import med /TREEVIEW_AUTO_COLLAPSE_REGEXP/ variablen. 
+ - Journaler inkluderes ved eksport.
+ - Bakgrunnsmodus: Støtter autoeksport med kommandolinjeparameterene /--file/, /--file-format/ og /--auto-export/.
+
 ** innsyn-v1.2.0-beta1
 *** Nye funksjoner
  - Støtte for standard ut mappe ved eksport, settes med konfigurasjonsvariablen /FIXED_REPORT_EXPORT_FOLDER/.
@@ -281,5 +288,10 @@ sudo apt install libquazip5-dev
 sudo apt install qttools5-dev-tools
 
 ** Windows
+#+BEGIN_SRC
+setup-win64.cmd
+qmake -tp vc
+#+END_SRC
+
 Her må man bygge poppler manuelt. 
 

nha-sip-convert.cmd

@@ -0,0 +1,24 @@
+setlocal
+set IN=%1
+echo %0
+echo %dirname%
+
+for %%F in (%0) do set dirname=%%~dpF
+
+; set PATH=%PATH%;%dirname%
+cd %dirname%
+
+
+set timestamp=%date%_%time:~0,-3%
+set timestamp=%timestamp::=_%
+set timestamp=%timestamp:.=_%
+set timestamp=%timestamp:/=_%
+set timestamp=%timestamp: =_%
+
+set OUT=export-%timestamp%.pdf
+
+set PLATFORM=-platform windows:dpiawareness=0
+
+insight %PLATFORM% --file %IN% --file-format "Norsk Helsearkiv SIP" --auto-export %OUT%
+
+pause

pdf/HocrConverter

@@ -0,0 +1 @@
+Subproject commit 25d84bafc17b4191d7492260e65fe5a17e266182

pdf/alto2hocr.cmd

@@ -3,8 +3,24 @@ setlocal
 set XSLT_TOOL="C:\Program Files\Saxonica\SaxonHE10.5N\bin\transform"
 set ALTO2HOCR_XSLT=alto__hocr.xsl
 
+set ERRORLOG=%2
+shift /2
+
+%XSLT_TOOL% -? > nul
+if not %ERRORLEVEL% equ 0 (
+    echo %0: %XSLT_TOOL% not in path >> %errorlog%
+    EXIT /B 1
+)
+
+if not exist %ALTO2HOCR_XSLT% (
+    echo %0: %ALTO2HOCR_XSLT% not in path >> %errorlog%
+    EXIT /B 1
+)
+
+
 if "%2" NEQ "" SET OUTHOCR="-o:%2" 
 
+echo %XSLT_TOOL% -s:%1 -xsl:%ALTO2HOCR_XSLT% %OUTHOCR% >> %errorlog%
 %XSLT_TOOL% -s:%1 -xsl:%ALTO2HOCR_XSLT% %OUTHOCR%
 
 

pdf/alto2pdf.cmd

@@ -1,9 +1,26 @@
-@echo off
 setlocal
 set IN_JPG=%1
 set IN_ALTO=%2
 set OUT=%3
+set ERRORLOG=%4
+
+python --help > nul
+if not %ERRORLEVEL% equ 0 (
+    echo %0: python not in path >> %ERRORLOG%
+    EXIT /B 1
+)
+
+set HOCR_SCRIPT=.\HocrConverter\HocrConverter.py
+if not exist %HOCR_SCRIPT% (
+    echo ERROR %HOCR_SCRIPT% not found >> %ERRORLOG%
+)
 
 rem set DEBUG_PARAMS=-b -v -t
 
-alto2hocr.cmd %IN_ALTO% | python .\HocrConverter\HocrConverter.py -i - -I -V -n -o %OUT% %DEBUG_PARAMS% %IN_JPG% 
+rem echo "alto2hocr.cmd %IN_ALTO% | python %HOCR_SCRIPT% -i - -I -V -n -o %OUT% %DEBUG_PARAMS% %IN_JPG%" >> %ERRORLOG%
+
+alto2hocr.cmd %IN_ALTO% %errorlog% | python %HOCR_SCRIPT% -i - -I -V -n -o %OUT% %DEBUG_PARAMS% %IN_JPG% >> %ERRORLOG% 2>&1
+if not %ERRORLEVEL% equ 0 (
+    echo %0: ERROR executing "alto2hocr.cmd %IN_ALTO% | python %HOCR_SCRIPT% -i - -I -V -n -o %OUT% %DEBUG_PARAMS% %IN_JPG%" >> %ERRORLOG%
+    exit /b 1
+)