I am confused by seemingly contradicting information in the PDF Standard (PDF 32000-1:2008).
To simplify, I assume no transparency complications at all, as I am confused enough as is, so alphas are all 1, and I have blending mode normal, and text knockout is not a topic.
The issue is text state (more precisely Tr, setting the text rendering mode) and text showing (Tj and friends) and how to reconcile these with the scope of what a SINGLE path is.
So these three things, A, B, and C, seem contradictory:
A. On p.247 (in section 9.3.6, under Table 106), it says (emphasis with caps is mine): "At the end of the text object, the accumulated glyph outlines, if any, shall be combined INTO A SINGLE PATH, treating the individual outlines as subpaths of that path and applying the nonzero winding number rule (...)."
B. According to p.113, Figure 9 ("Graphics Objects"), the allowed operators within a text object include: "Color" (e.g. a change of fill color), "Text state" (so e.g. changing Tr, the text rendering mode), and "Text-showing" (e.g. (Hello) Tj).
According to B, the following would then be a valid text object:
C. On p. 246, in the middle of the page it says (emphasis in caps is mine): "In [text rendering] modes that perform both filling and stroking, the effect shall be as if each glyph outline where filled and then stroked in separate operations. If any of the glyphs overlap, the result shall be equivalent to filling and stroking them ONE AT A TIME, producing the appearance of stacked opaque glyphs, rather then first filling and stroking them all at once".
So, according to A, all the glyphs (in my example in B) in "Red hello filled" and "Blue hello filled then stroked" must be considered as A SINGLE path, and the individual glyphs are subpaths...
My understanding is that ONE path can be filled with ONE color (I do not mean gradients of course, I am talking about solid colors, like the red and the blue in my example in B), is this not so? And I can only apply 1 text rendering mode to it, right? How does it make sense then that I can change color and text rendering mode WITHIN the text object, which B tells me, if according to A, I am dealing with a SINGLE path?
And according to C, even each glpyh is to be considered as a SINGLE path as we paint one after, so on top of, the other, which is exactely the concept of PATH. I can not fill ONE PATH "one at a time"...
So, bottomline question, what is the scope of a path in a PDF text object?
Thank you very much for help!
If I understand the spec correctly, you misinterpret the term "the accumulated glyph outlines". It does not refer to all the outline paths of the shown glyphs and their rendering as you assume. Instead it refers only to the outlines that are added to path for clipping if any!
Thus, all the text is filled and/or stroked immediately in the text object according to the state at the time its text showing instruction is evaluated, one glyph at a time. Consequentially, text drawn by different text showing instructions may have different colors and different other characteristics and later glyphs may overlap earlier ones.
Eventually, at the end of a text object all the outlines of the text drawn with text rendering mode 4..7 in the object - if any - are combined in that single path A talks about and used to intersect the previous clipping path.
Related
I needed to convert some PDF back to text. I tried many soft and online tools and result was always mediocre.
Why is it so difficult technically speaking ?
Let's not assume you are talking about PDFs which merely wrap some bitmap image because it should be clear that in that case you can only resort to OCR with all its restrictions.
Let's instead assume that text is drawn in the PDF at hand.
What is drawn on a PDF page is determined by a sequence of instructions in the content stream of that page. "Text is drawn" on a page means that among those instructions there are some setting the font to use by the instructions to come, some setting the text position and direction to use by the instructions to come, and some actually drawing text given by "string arguments".
Text extraction is the task of taking the sequence of instructions from a content stream and instead of drawing the text as indicated by the font and position setting instructions, to export it in a sensible order using a standard encoding, usually the encoding of the character type of the used programming language / platform.
The first problem is to understand the encoding of the string arguments of those text drawing instructions:
each font can have its own encoding; to extract the text one cannot simply ignore everything but the instructions drawing text and concatenate their string contents, you always have to take the current font into account (some extremely simple text extractors ignore this and, therefore, fail pretty often to return something sensible);
there are a large number of predefined encodings, some reminding of encodings you know, e.g. WinAnsiEncoding, many you likely don't know, e.g. Add-RKSJ-H; these encodings may use a constant number of bytes per glyph or they may be mixed-multibyte; so a text extractor must support very many encodings to start with;
encodings also may be completely ad-hoc and arbitrary; in particular in case of embedded subset fonts one often sees ad-hoc encodings generated by dealing out character codes from some starting value whenever one is needed; i.e. the first glyph in a given font used on a page is given the starting value as code, the next, different glyph is given the starting value plus one, the next, different one the starting value plus two, etc; "Hello World" and a starting value of 48 (ASCII value of '0') would result in "01223453627"; these fonts may contain a mapping to Unicode but they are not required to.
The next problem is to make sense out of the order of the strings:
the string drawing instructions may occur in an arbitrary order, e.g "Hello" might be drawn "lo" first, then after moving back "el", then after again moving back "H"; to extract the text one cannot ignore text positioning instructions and simply concatenate text strings, you always have to take the current position into account (some simple text extractors ignore this and, therefore, can fail to return something sensible);
multi-columnar text may present a difficulty, text may be drawn line by line, e.g. first the text of the top line of the first column, then the top line of the second column, then the second line of the first column, then the second line of the second column, etc.; there need not be any hints in the PDF that the text is multi-columnar.
Another problem is to recognize formatting or styling artifacts:
spaces between words need not be created by drawing a space glyph, it may also be done by text position changing instructions; text extractors not trying to recognize gaps created by text positioning instructions may return a result without spaces; on the other hand the same technique can be used to draw adjacent glyphs at an optimal distance, aka kerning; text extractors trying to recognize gaps created by text positioning instructions may falsely return spaces where there should be none;
sometimes selected words are printed s p a c e d o u t for extra emphasis; in the extracted text these gaps might be presented as space characters which automatic postprocessing of the text may see as word separators;
usually for bold text one uses a different, bold font program; if that is not at hand, people sometimes get creative and emulate bold by printing the same text twice with a minute offset; with a slightly larger offset (or a different transformation) and a different color a shadow effect can be emulated; if the text extractor does not try to recognize this, you end up having some duplicate characters in the output.
More problems arise due to incomplete or wrong extra information:
ToUnicode maps of fonts (optional maps from character code to Unicode) may be incomplete or contain errors; there e.g. are many questions here on stack overflow dealing with incorrect ToUnicode maps for Indian writings; the text extraction results reflect these errors;
there even are PDFs with contradictory information, e.g. with an error in the ToUnicode map but the correct information in an ActualText entry; this is used by some PDF creators to allow correct copy&paste from some programs (preferring an ActualText entry in such a situation) while injecting errors in the output of other programs (preferring ToUnicode information then).
Yet another problem arises if you expect the text extractor to extract only text eventually visible in the page:
text may be drawn outside the current clipping area or outside the visible page area; text extractors need to keep these in mind;
text may be drawn using the rendering mode "invisible"; text extractors have to keep an eye on the rendering mode;
text may be drawn using the same color as the background; to recognize this, a text extractor can not only look at the current instruction and a few graphics state details, it has to take into account anything drawn beforehand in the location of the text;
text may be drawn as a clip path; to recognize whether this text is visible in the end, a text extractor must keep track of what is drawn in the text area as long as the clip path is active;
text may be covered by something else later; a text extractor must drop recognized text in such a case; but depending on blend modes and transparency settings these coverings might or might not allow the text to shine through; thus, for a correct result the text extractor must for each glyph keep track of the color its drawn with, the color of the backdrop, and what all those spiffy effects do with those colors later on; and of course, both glyph color and backdrop color can be interesting, e.g. some shading colors; and the color spaces involved may differ, requiring one to convert back and forth between color spaces; and so on.
Furthermore, text may be drawn where text extractors usually don't look:
some tools hide text from text extraction by putting it into a pattern and filling the page area with that pattern;
similarly there are type 3 fonts; each character in a type 3 font is represented by its own content stream; thus, a tool can draw all text in the content stream of a single type 3 font glyph and then draw that glyph on the page.
...
You surely have meanwhile gotten an idea why text extraction results can be less than optimal. And be assured, the list above is not complete, there still are more complications for text extraction.
I am using PDFBox 2.0.8, I am able to identify the text position of character, but it does not help to determine if there is an overlapping text. Example: "in excess o$f1 ,000.96". As you can see the $ symbol overlaps on "of", but how to predict that there is an overlap?
I have found a workable solution myself to the problem that I have reported. While processing the List, I have added following condition that verifies if there is an overlapping of text. This solution may not work for all kinds of pdf documents, some additional work might need to be done:
if(textPositions.indexOf(position)+1<textPositions.size()){ if(position.contains(textPositions.get(textPositions.indexOf(position)+1))){overlappingTextFlag = true}}
public boolean contains(TextPosition tp2)
Determine if this TextPosition logically contains another (i.e. they overlap and should be rendered on top of each other).
Parameters:
tp2 - The other TestPosition to compare against
Returns:
True if tp2 is contained in the bounding box of this text.
I have run into a problem with iText 7 where diacritic marks are painted on top of one another instead of stacking properly when multiple marks are used on a single character. Is there a setting that makes them appear correctly, or is this a bug in iText 7? Any help greatly appreciated. This can be observed if you create a text object in your PDF like below. Obviously, replace the relevant bit with an actual font object, rather than than what I have in there.
new Text("ḗ and ṓ are characters that display incorrectly").setFont(<UNICODE COMPATIBLE FONT LIKE CHARIS>);
While Bruno and Benoit correctly pointed out that for advanced typography stuff like stacking diacritical marks you need pdfCalligraph module, there is a workaround you can try at your own risk. If your combinations of base glyph and diacritics are real, meaning they occur in real texts in some languages or some other known contexts, then such combinations are most probably present in Unicode and have their own number associated with them. For instance, in text you provided, they are 0xU1E17 and 0x1E53 Unicode characters. Some fonts may contain such glyphs, so that there is a second option to showing base glyph and stacking diacritics: showing combined glyphs. For example, ArialUni shipped with Windows does contain the above mentioned glyphs.
To try this approach, you would need the following code for composing known Unicode base glyph + diacritics combinations into single glyphs:
String originalStr = "ḗ and ṓ are characters that display incorrectly";
String normalizedStr = java.text.Normalizer.normalize(originalStr, Normalizer.Form.NFC);
new Text(normalizedStr); // Use this normalized Text instance
The result that I got with ArialUni:
But again as I mentioned do it at your own risk because it will only work if there are necessary combinations present both in Unicode and font. For correct rendering you still should use pdfCalligraph.
Does anyone know if the method getFontSize in TextPosition always returns one and should I only use getFontSizeInPt to get the size of the font?
The problem I have is that getFontSizeInPt sometimes returns different values for the same sized text (I got 12 and 11 return for text in the same paragraph with the same size.
Does anyone know if the method getFontSize in TextPosition always returns one
It does not always return one.
Please be aware that in the PDF page content descriptions there are several settings which all influence the final text size:
the font size parameter of the font selecting operator Tf:
the text matrix set by the operator Tm;
the current transformation matrix set by the operator cm;
the UserUnit setting of the PDF page.
The final text size is the first value scaled by the text matrix, scaled again by the transformation matrix, and scaled once more by the user unit value.
(Actually there even are some more factors. E.g. if one uses rendering mode 2, fill & stroke, for a faux bold effect, this slightly increases the size, too.)
TextPosition.getFontSize returns the first value only.
TextPosition.getFontSizeInPt returns something like the first value scaled by the matrices. (something like because at first glance there seems to be another influence in it.)
Different PDF creators use these influences in different ways:
Some PDF creators use only the first value to set the font size and use the matrices only for operations not changing the effective font size, e.g. rotations.
Some PDF creators set the first value to 1 and scale using the matrices.
Some PDF creators fall inbetween and use both the first value and the scaling operations.
Thus, your PDFs seem to be created by software using the second way.
getFontSizeInPt sometimes returns different values for the same sized text (I got 12 and 11 return for text in the same paragraph with the same size.
Could you share a sample PDF with that issue? As mentioned above, at first glance there seem to be additional influences which might be incorrect. But there also might be something special about your PDF.
Say I have a PDF file that contains one or more embedded fonts. Here's my understanding of how a single character of text is rendered:
First, determine which font the character uses.
Use the font's "cmap," embedded in the PDF, to determine the font's glyph name for the given character. For example, the character '&' in PDF text might map to a glyph that the font internally calls 'ampersand'.
Use the font's "glyf" table to determine the bounding box / drawing instructions for the glyph name.
Here's my question: is a PDF cmap generally consistent? Put another way, if I encounter the character "&" in a PDF, can I be assured that the cmap will always map "&" to the ampersand glyph? Or does some PDF-generation software create its own arbitrary mapping between character codes and glyph names (which would be rather evil and possibly break in-PDF searching and text selection)?
Of course I realize it's possible for the cmap to use an unintuitive mapping -- I guess I'm asking, does this actually happen in the Real World?
My specific use-case is in the world of music fonts. I'm analyzing characters in a PDF to determine which music glyph each one represents (e.g., treble clef, notehead, etc.). I want to know how confident I can be that the combination of font name and character code will always result in the same glyph. For example, if I know the font name is "Opus" and the glyph is "#", can I assume that will always be mapped to the treble clef glyph? Or do I have to analyze the glyph's metrics to make sure it's actually a treble clef?
It differs from one PDF creator to another.
A fairly common method (alas!) is "order encountered", where the first character in the text stream gets mapped to \001, the next to \002 and so on. So the text "Hello" would be encoded as \001\002\003\003\004.
I want to know how confident I can be that the combination of font name and character code will always result in the same glyph.
In a single PDF document, if the same font object is used in different contexts, it will be true -- the mapping is defined inside the font object. If you encounter another font object that uses the same font but it points to another font stream (i.e., the font subset is embedded twice), then it may not be true. Each subset may have an encoding of its own.
Only if the font object contains a /ToUnicode mapping, you can be confident that values map to the correct characters.