fix many internal links

Fixed a number of internal links that need to
go to the GitHub site.

Ran `make www` to perform the above.

2019/ciura/README.md

@@ -132,14 +132,14 @@ is too long, and 2 if the command-line arguments are missing.
 
 ### Helper scripts:
 
-- [getwords.sh](getwords.sh)
+- [getwords.sh](%%REPO_URL%%/2019/ciura/getwords.sh)
 
     `./getwords.sh LANGUAGE_CODE`
 
     outputs a sorted list of unique words in a given language, one per line.  It
     requires `aspell` with a dictionary for _LANGUAGE_CODE_.
 
-- [wrapprog.py](wrapprog.py)
+- [wrapprog.py](%%REPO_URL%%/2019/ciura/wrapprog.py)
 
     `./wrapprog.py LETTERS [N]`
 
@@ -148,8 +148,9 @@ is too long, and 2 if the command-line arguments are missing.
     it finishes the job faster than `prog` when there are millions of perfect
     pangrams. It requires a UTF-8 locale.
 
-- [de.sh](de.sh), [en.sh](en.sh), [fr.sh](fr.sh), [it.sh](it.sh),
-[pl.sh](pl.sh), and [ru.sh](ru.sh) output German,
+- [de.sh](%%REPO_URL%%/2019/ciura/de.sh), [en.sh](%%REPO_URL%%/2019/ciura/en.sh),
+[fr.sh](%%REPO_URL%%/2019/ciura/fr.sh), [it.sh](%%REPO_URL%%/2019/ciura/it.sh),
+[pl.sh](%%REPO_URL%%/2019/ciura/pl.sh), and [ru.sh](%%REPO_URL%%/2019/ciura/ru.sh) output German,
 English, French, Italian, Polish, and Russian perfect pangrams, respectively.
 They require a UTF-8 locale.
 
@@ -162,7 +163,7 @@ among the programs that won IOCCC in the past), and _obscure bugs_ (see
 the questions about Linux core dumps below).
 
 To a casual eye, this entry may look similar to
-[2005/klausler](../../2005/klausler/index.html).
+[2005/klausler](%%REPO_URL%%/2005/klausler/index.html).
 However, `prog` is internationalized and way faster. On my machine,
 `./klausler abcdefghijklmnopqrstuvwxyz` printed only 3,965 perfect
 pangrams until I killed it after 144 hours running, while the
@@ -173,7 +174,7 @@ the example from section [What is this?](#wit) finished in 1.35
 seconds. On the other hand, `prog` can only output series of words
 with non-repeating characters, unlike `klausler`.
 
-In contrast to [prog.alt.c](prog.alt.c), [prog.c](prog.c) does not call
+In contrast to [prog.alt.c](%%REPO_URL%%/2019/ciura/prog.alt.c), [prog.c](%%REPO_URL%%/2019/ciura/prog.c) does not call
 `fflush(stdout)` after outputting each line, thus running faster.
 I am grateful to Witold Jarnicki for suggesting this change.
 The original `prog` finished the example above in 100 minutes
@@ -201,8 +202,8 @@ by without `goto M` and `goto H`, I challenge the adherents of
 structured programming to refactor `goto T`, which jumps back into a
 nested `if` block.
 
-With the supplied [Makefile](Makefile), both `gcc` and `clang` compile
-[prog.c](prog.c) without warnings in C11 and C99 mode. For a clean
+With the supplied [Makefile](%%REPO_URL%%/2019/ciura/Makefile), both `gcc` and `clang` compile
+[prog.c](%%REPO_URL%%/2019/ciura/prog.c) without warnings in C11 and C99 mode. For a clean
 compilation with `gcc -std=c90`, add `-Wno-format` to
 `CSILENCE` in `Makefile`.
 

2019/ciura/index.html

@@ -451,18 +451,19 @@ unlimited. The characters in <code>LETTERS</code> must not repeat. The length of
 is too long, and 2 if the command-line arguments are missing.</p>
 <h3 id="helper-scripts">Helper scripts:</h3>
 <ul>
-<li><p><a href="getwords.sh">getwords.sh</a></p>
+<li><p><a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/getwords.sh">getwords.sh</a></p>
 <p><code>./getwords.sh LANGUAGE_CODE</code></p>
 <p>outputs a sorted list of unique words in a given language, one per line. It
 requires <code>aspell</code> with a dictionary for <em>LANGUAGE_CODE</em>.</p></li>
-<li><p><a href="wrapprog.py">wrapprog.py</a></p>
+<li><p><a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/wrapprog.py">wrapprog.py</a></p>
 <p><code>./wrapprog.py LETTERS [N]</code></p>
 <p>works similarly to <code>./prog LETTERS [N]</code> except that it merges pangrams
 before passing them to <code>prog</code> and expands them in <code>prog</code>’s output. This way,
 it finishes the job faster than <code>prog</code> when there are millions of perfect
 pangrams. It requires a UTF-8 locale.</p></li>
-<li><p><a href="de.sh">de.sh</a>, <a href="en.sh">en.sh</a>, <a href="fr.sh">fr.sh</a>, <a href="it.sh">it.sh</a>,
-<a href="pl.sh">pl.sh</a>, and <a href="ru.sh">ru.sh</a> output German,
+<li><p><a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/de.sh">de.sh</a>, <a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/en.sh">en.sh</a>,
+<a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/fr.sh">fr.sh</a>, <a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/it.sh">it.sh</a>,
+<a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/pl.sh">pl.sh</a>, and <a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/ru.sh">ru.sh</a> output German,
 English, French, Italian, Polish, and Russian perfect pangrams, respectively.
 They require a UTF-8 locale.</p></li>
 </ul>
@@ -472,7 +473,7 @@ They require a UTF-8 locale.</p></li>
 among the programs that won IOCCC in the past), and <em>obscure bugs</em> (see
 the questions about Linux core dumps below).</p>
 <p>To a casual eye, this entry may look similar to
-<a href="../../2005/klausler/index.html">2005/klausler</a>.
+<a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2005/klausler/index.html">2005/klausler</a>.
 However, <code>prog</code> is internationalized and way faster. On my machine,
 <code>./klausler abcdefghijklmnopqrstuvwxyz</code> printed only 3,965 perfect
 pangrams until I killed it after 144 hours running, while the
@@ -482,7 +483,7 @@ finished in 62 minutes, reporting 1,640,569,315 perfect pangrams, and
 the example from section <a href="#wit">What is this?</a> finished in 1.35
 seconds. On the other hand, <code>prog</code> can only output series of words
 with non-repeating characters, unlike <code>klausler</code>.</p>
-<p>In contrast to <a href="prog.alt.c">prog.alt.c</a>, <a href="prog.c">prog.c</a> does not call
+<p>In contrast to <a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/prog.alt.c">prog.alt.c</a>, <a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/prog.c">prog.c</a> does not call
 <code>fflush(stdout)</code> after outputting each line, thus running faster.
 I am grateful to Witold Jarnicki for suggesting this change.
 The original <code>prog</code> finished the example above in 100 minutes
@@ -504,8 +505,8 @@ have no perfect pangrams.</p>
 by without <code>goto M</code> and <code>goto H</code>, I challenge the adherents of
 structured programming to refactor <code>goto T</code>, which jumps back into a
 nested <code>if</code> block.</p>
-<p>With the supplied <a href="Makefile">Makefile</a>, both <code>gcc</code> and <code>clang</code> compile
-<a href="prog.c">prog.c</a> without warnings in C11 and C99 mode. For a clean
+<p>With the supplied <a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/Makefile">Makefile</a>, both <code>gcc</code> and <code>clang</code> compile
+<a href="https://github.com/ioccc-src/temp-test-ioccc/blob/master/2019/ciura/prog.c">prog.c</a> without warnings in C11 and C99 mode. For a clean
 compilation with <code>gcc -std=c90</code>, add <code>-Wno-format</code> to
 <code>CSILENCE</code> in <code>Makefile</code>.</p>
 <!--