See this on atom.io
Given the nature of LaTeX, it is practically impossible to get an accurate word count for any non-trivial document. However, this package provides several methods of estimating the word count using tools that come with a TeX Live distribution, such as
Most of the provided commands use
TeXcount, as it is the quicker and more lenient program. However, it is likely to underestimate the wordcount most of the time. The one command that uses
wordcount.tex only works on the entire document and requires generating a log file that grows larger as the document gets longer. This one will likely overestimate the word count, so (to reiterate the earlier point) these methods are approximations only and you as the user must decide how to interpret the results.
texcountcommand line program. This comes with a standard TeX Live installation. I believe it also comes with MiKTeX, but cannot confirm.
Document wordcount (more accurate)command only works on a UNIX shell, as it currently uses commands such as
grep. Most Windows users will be unable to use this command for now.
/Users/username/my TeX files/main.texwill not work.
Document wordcount: counts entire document, from start of root file, including all
Document wordcount (more accurate): more accurate version of Document wordcount, but takes longer and requires generation of a log file containing a line for every character in the output.
File wordcount: counts only current file, and does not count
\inputfiles (the only one that doesn't).
Section wordcount: counts the current section, including all
Selection wordcount: counts the current selection, including
This will count using TeXcount, which uses a set of LaTeX specific rules to better estimate the word count. It's quite fast, as it does not compile the document, so useful if you want a rough estimate of the word count. Note that it will not know any macro definitions, so take this into consideration if you use macros that expand into a group of words. It will probably underestimate the word count in most cases. See it's homepage here.
Forgive the name. (this does not work on files with spaces in the path; see top note)
This command runs a different program, and actually compiles your document. The output should be a file called
wordcount.log located in the same directory as the root file (the one that gets compiled; magic comments are followed). Basically, it redefines several TeX commands to force the log file to include each character and word. A simple search command then finds the number of characters and words, and prints the result.
This method is more accurate than TeXcount; it 'knows' macro definitions, so (for example) it will correctly give the number of words inserted by the
However, it makes no distinction between the type of any text. Tables, captions, math formulas (each individual term; even superscripts are counted), code (in a
minted block), etc. will all be included in the count.
Back to TeXcount, this command will run it on the currently open file (not the root one). It will also ignore any
\include, etc. commands that would be honoured by
First, the text in the current section is gathered. This section is determined by looking back from the cursor position to the nearest section command, where section command is one of the following:
\subparagraph. However, only section commands of the level set by maximum section depth or lower (shallower?) will be recognised. When found, the section start (that will eventually be passed to
texcount) is set to the beginning of that section command.
Next, it will then look for the next section command of same or lower level as the starting one. When found, it will set the end of the section range to just before this section command.
Finally, it copies the text in this range and pastes it into a temporary file created by the npm
tmp package. It then runs
texcount on this temporary file, but set up so it thinks it's in the same directory as the original file, ensuring relative file paths still work (as
\input files are included if they are within the section).
Similar to Section wordcount, this command will count the words in a selected region. Specifically, it will create a temporary file in the system temp directory and write the selected text to it. It will then run
texcount on this temporary file, specifying the directory as that of the original file. This way,
\input statements (and others) will still work if they are completely within the selection. TeXcount will not see the outside of the selection, so if the selection cuts into a macro it may not behave as expected.
In both this case and the above, the temporary file is deleted as soon as the results are returned.
The shell escape flag is generally used to allow LaTeX to execute arbitrary code. This can be useful for packages such as
minted, where the syntax colours are determined using an external program.
This option is not necessary for TeXcount, as it works by parsing the source. However,
wordcount.tex requires the document be compiled, so any packages that require shell escape need this flag enabled.
When counting the words in a section, this setting determines the deepest section command to look for. For example, consider the following document layout
A setting of
subsection will cause the count when the cursor is at (1) to be for the entire
S:2, the count at (2) to be between
SS:1-2, and the count at (3) to be between
A setting of
section will cause all three locations to count the same area (
S:2), because the deeper
subsection commands will be ignored.
Good catch. Let us know what about this package looks wrong to you, and we'll investigate right away.