x_fsize always 0 in hocr-output since latest commit

[hnesk]

Environment

• Tesseract Version: 4.0.0-86-gbee8
• Commit Number: ad40131
• Platform: Linux jk-XPS-13 4.18.0-12-generic Issue 1353: Patch for /training/tessopt.cpp #13-Ubuntu SMP Wed Nov 14 15:17:05 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

Current Behavior:

With current git master version:
$ ~/bin/tesseract -v
tesseract 4.0.0-86-gbee8
leptonica-1.76.0
libgif 5.1.4 : libjpeg 8d (libjpeg-turbo 1.5.2) : libpng 1.6.34 : libtiff 4.0.9 : zlib 1.2.11 : libwebp 0.6.1 : libopenjp2 2.3.0
Found AVX2
Found AVX
Found SSE

$ ~/bin/tesseract --psm 3 --oem 0 -c tessedit_create_hocr=1 -c hocr_font_info=1 -l deu page.tif page-4.0.0-86-gbee8

In page-4.0.0-86-gbee8.hocr: "x_fsize 0" and "x_font" is missing completly:
<span class='ocrx_word' id='word_1_1' title='bbox 253 248 365 292; x_wconf 89; x_fsize 0'>rung</span>

Expected Behavior:

With the current ubuntu version 4.0.0-beta.3-249-g607e:
$ tesseract -v
tesseract 4.0.0-beta.3-249-g607e
leptonica-1.76.0
libgif 5.1.4 : libjpeg 8d (libjpeg-turbo 1.5.2) : libpng 1.6.34 : libtiff 4.0.9 : zlib 1.2.11 : libwebp 0.6.1 : libopenjp2 2.3.0
Found AVX2
Found AVX
Found SSE
$ tesseract --psm 3 --oem 0 -c tessedit_create_hocr=1 -c hocr_font_info=1 -l deu page.tif page-4.0.0-beta.3-249-g607e

In page-4.0.0-beta.3-249-g607e.hocr: "x_fsize" and "x_font" have sensible values:
<span class='ocrx_word' id='word_1_1' title='bbox 253 248 365 292; x_wconf 89; x_font Times_New_Roman; x_fsize 56'>rung</span>

Suggested Fix:

review c9e85ab or ad40131 one of these commits fixed too much ;)

[hnesk]

Testcase:
page.zip

[amitdo]

c9e85ab

if (it_->word()) {

should be:
if (it_->word() == nullptr) {

[stweil]

@hnesk and @amitdo, thank you for your reports and sorry for the regression. It is fixed now with commit 2c044df.

[stweil]

x_fsize is not written by default because the hocr config file sets hocr_font_info 0, so I assume most users won't notice the bug.

[hnesk]

Great! Works as expected with 2c044df