package/tesseract-ocr/tesseract-ocr.mk

   1 ################################################################################
   2 #
   3 # tesseract-ocr
   4 #
   5 ################################################################################
   6
   7 TESSERACT_OCR_VERSION = 3.05.00
   8 TESSERACT_OCR_DATA_VERSION = 3.04.00
   9 TESSERACT_OCR_SITE = $(call github,tesseract-ocr,tesseract,$(TESSERACT_OCR_VERSION))
  10 TESSERACT_OCR_LICENSE = Apache-2.0
  11 TESSERACT_OCR_LICENSE_FILES = COPYING
  12
  13 # Source from github, no configure script provided
  14 TESSERACT_OCR_AUTORECONF = YES
  15
  16 # cairo, pango and icu are optional dependencies, but only needed for
  17 # building training tools, which are only built explicitly with "make
  18 # training", which is not done by this package.
  19 TESSERACT_OCR_DEPENDENCIES = leptonica host-pkgconf
  20 TESSERACT_OCR_INSTALL_STAGING = YES
  21 TESSERACT_OCR_CONF_ENV = \
  22         LIBLEPT_HEADERSDIR=$(STAGING_DIR)/usr/include/leptonica
  23 TESSERACT_OCF_CONF_OPTS = \
  24         --disable-opencl
  25
  26 # Language data files download
  27 ifeq ($(BR2_PACKAGE_TESSERACT_OCR_LANG_ENG),y)
  28 TESSERACT_OCR_DATA_FILES += eng.traineddata
  29 endif
  30
  31 ifeq ($(BR2_PACKAGE_TESSERACT_OCR_LANG_FRA),y)
  32 TESSERACT_OCR_DATA_FILES += fra.traineddata
  33 endif
  34
  35 ifeq ($(BR2_PACKAGE_TESSERACT_OCR_LANG_DEU),y)
  36 TESSERACT_OCR_DATA_FILES += deu.traineddata
  37 endif
  38
  39 ifeq ($(BR2_PACKAGE_TESSERACT_OCR_LANG_SPA),y)
  40 TESSERACT_OCR_DATA_FILES += spa.traineddata
  41 endif
  42
  43 ifeq ($(BR2_PACKAGE_TESSERACT_OCR_LANG_CHI_SIM),y)
  44 TESSERACT_OCR_DATA_FILES += chi_sim.traineddata
  45 endif
  46
  47 ifeq ($(BR2_PACKAGE_TESSERACT_OCR_LANG_CHI_TRA),y)
  48 TESSERACT_OCR_DATA_FILES += chi_tra.traineddata
  49 endif
  50
  51 TESSERACT_OCR_EXTRA_DOWNLOADS = \
  52         $(addprefix https://github.com/tesseract-ocr/tessdata/raw/$(TESSERACT_OCR_DATA_VERSION)/,\
  53                 $(TESSERACT_OCR_DATA_FILES))
  54
  55 define TESSERACT_OCR_PRECONFIGURE
  56         # Autoreconf step fails due to missing m4 directory
  57         mkdir -p $(@D)/m4
  58 endef
  59
  60 TESSERACT_OCR_PRE_CONFIGURE_HOOKS += TESSERACT_OCR_PRECONFIGURE
  61
  62 # Language data files installation
  63 define TESSERACT_OCR_INSTALL_LANG_DATA
  64         $(foreach langfile,$(TESSERACT_OCR_DATA_FILES), \
  65                 $(INSTALL) -D -m 0644 $(DL_DIR)/$(langfile) \
  66                         $(TARGET_DIR)/usr/share/tessdata/$(langfile)
  67         )
  68 endef
  69
  70 TESSERACT_OCR_POST_INSTALL_TARGET_HOOKS += TESSERACT_OCR_INSTALL_LANG_DATA
  71
  72 $(eval $(autotools-package))