[text_format] add shell script

NEWS.md

@@ -123,6 +123,8 @@ Features:
   table that contains a JSON object with the top 5 values
   for the fields extracted from the log message.
 * Added Nextcloud log format from Adam Monsen.
+* Added GitHub Event Log format for files from gharchive.org.
+  It makes a good example of a JSON-Lines format.
 
 Bug Fixes:
 * Binary data piped into stdin should now be treated the same

src/text_format.cc

@@ -67,6 +67,7 @@ detect_text_format(string_fragment sf,
     static const auto MAKEFILE_STEM = ghc::filesystem::path("Makefile");
     static const auto MD_EXT = ghc::filesystem::path(".md");
     static const auto MARKDOWN_EXT = ghc::filesystem::path(".markdown");
+    static const auto SH_EXT = ghc::filesystem::path(".sh");
 
     static const auto DIFF_MATCHERS = lnav::pcre2pp::code::from_const(
         R"(^--- .*\n\+\+\+ .*\n)", PCRE2_MULTILINE);
@@ -129,6 +130,9 @@ detect_text_format(string_fragment sf,
         ")",
         PCRE2_MULTILINE | PCRE2_CASELESS);
 
+    static const auto SH_MATCHERS
+        = lnav::pcre2pp::code::from_const("^#!.+sh\\b", PCRE2_MULTILINE);
+
     if (path) {
         while (FILTER_EXTS.count(path->extension()) > 0) {
             path = path->stem();
@@ -171,6 +175,10 @@ detect_text_format(string_fragment sf,
         if (stem == MAKEFILE_STEM) {
             return text_format_t::TF_MAKEFILE;
         }
+
+        if (stem == SH_EXT) {
+            return text_format_t::TF_SHELL_SCRIPT;
+        }
     }
 
     {
@@ -186,6 +194,10 @@ detect_text_format(string_fragment sf,
         return text_format_t::TF_DIFF;
     }
 
+    if (SH_MATCHERS.find_in(sf).ignore_error()) {
+        return text_format_t::TF_SHELL_SCRIPT;
+    }
+
     if (MAN_MATCHERS.find_in(sf).ignore_error()) {
         return text_format_t::TF_MAN;
     }

src/text_format.hh

@@ -57,6 +57,7 @@ enum class text_format_t {
     TF_YAML,
     TF_TOML,
     TF_DIFF,
+    TF_SHELL_SCRIPT,
 };
 
 namespace fmt {
@@ -115,6 +116,9 @@ struct formatter<text_format_t> : formatter<string_view> {
             case text_format_t::TF_DIFF:
                 name = "text/x-diff";
                 break;
+            case text_format_t::TF_SHELL_SCRIPT:
+                name = "text/x-shellscript";
+                break;
         }
         return formatter<string_view>::format(name, ctx);
     }

src/textfile_highlighters.cc

@@ -455,6 +455,11 @@ setup_highlights(highlight_map_t& hm)
                             ")"))
               .with_nestable(false)
               .with_role(role_t::VCR_VARIABLE);
+    hm[{highlight_source_t::INTERNAL, "yaml.var"}]
+        = highlighter(xpcre_compile("^\\s*(?:- )?[a-zA-Z_\\-]+:(?:\\s+|$)"))
+              .with_nestable(false)
+              .with_text_format(text_format_t::TF_YAML)
+              .with_role(role_t::VCR_VARIABLE);
     hm[{highlight_source_t::INTERNAL, "rust.sym"}]
         = highlighter(xpcre_compile("\\b[A-Z_][A-Z0-9_]+\\b"))
               .with_nestable(false)
@@ -479,11 +484,57 @@ setup_highlights(highlight_map_t& hm)
               .with_text_format(text_format_t::TF_C_LIKE)
               .with_text_format(text_format_t::TF_JAVA)
               .with_role(role_t::VCR_KEYWORD);
+    hm[{highlight_source_t::INTERNAL, "shell"}]
+        = highlighter(xpcre_compile("(?:"
+                                    "\\bbreak\\b|"
+                                    "\\bcase\\b|"
+                                    "\\bcd\\b|"
+                                    "\\bcontinue\\b|"
+                                    "\\bdeclare\\b|"
+                                    "\\bdefault\\b|"
+                                    "\\bdo\\b|"
+                                    "\\bdone\\b|"
+                                    "\\becho\\b|"
+                                    "\\belif\\b|"
+                                    "\\belse\\b|"
+                                    "\\besac\\b|"
+                                    "\\beval\\b|"
+                                    "\\bexit\\b|"
+                                    "\\bexport\\b|"
+                                    "\\bfalse\\b|"
+                                    "\\bfi\\b|"
+                                    "\\bfor\\b|"
+                                    "\\bfunction\\b|"
+                                    "\\bif\\b|"
+                                    "\\bin\\b|"
+                                    "\\blocal\\b|"
+                                    "\\bprintf\\b|"
+                                    "\\bpwd\\b|"
+                                    "\\bread\\b|"
+                                    "\\breadonly\\b|"
+                                    "\\breturn\\b|"
+                                    "\\bset\\b|"
+                                    "\\bshift\\b|"
+                                    "\\bsource\\b|"
+                                    "\\btest\\b|"
+                                    "\\bthen\\b|"
+                                    "\\btrap\\b|"
+                                    "\\btrue\\b|"
+                                    "\\bunset\\b|"
+                                    "\\bunsetenv\\b|"
+                                    "\\buntil\\b|"
+                                    "\\bwhich\\b|"
+                                    "\\bwhile\\b"
+                                    ")"))
+              .with_nestable(false)
+              .with_text_format(text_format_t::TF_SHELL_SCRIPT)
+              .with_role(role_t::VCR_KEYWORD);
     hm[{highlight_source_t::INTERNAL, "num"}]
         = highlighter(xpcre_compile(R"(\b-?(?:\d+|0x[a-zA-Z0-9]+)\b)"))
               .with_nestable(false)
               .with_text_format(text_format_t::TF_C_LIKE)
               .with_text_format(text_format_t::TF_JAVA)
+              .with_text_format(text_format_t::TF_YAML)
               .with_role(role_t::VCR_NUMBER);
     hm[{highlight_source_t::INTERNAL, "fun"}]
         = highlighter(xpcre_compile(R"((\w+)\()"))

test/expected/expected.am

@@ -1194,6 +1194,8 @@ EXPECTED_FILES = \
     $(srcdir)/%reldir%/test_sql_xml_func.sh_fefeb387ae14d4171225ea06cbbff3ec43990cf0.out \
     $(srcdir)/%reldir%/test_sql_yaml_func.sh_dc189d02e8979b7ed245d5d750f68b9965984699.err \
     $(srcdir)/%reldir%/test_sql_yaml_func.sh_dc189d02e8979b7ed245d5d750f68b9965984699.out \
+    $(srcdir)/%reldir%/test_text_file.sh_0bba304f34ae07c4fa9e91e0b42f5fe98654a6a8.err \
+    $(srcdir)/%reldir%/test_text_file.sh_0bba304f34ae07c4fa9e91e0b42f5fe98654a6a8.out \
     $(srcdir)/%reldir%/test_text_file.sh_1ce4056d72b871f8bb844c86aade2a9b1da58030.err \
     $(srcdir)/%reldir%/test_text_file.sh_1ce4056d72b871f8bb844c86aade2a9b1da58030.out \
     $(srcdir)/%reldir%/test_text_file.sh_4226123565a53b4e3f80e602c1f294721e8e07bf.err \

test/expected/test_text_file.sh_0bba304f34ae07c4fa9e91e0b42f5fe98654a6a8.out

@@ -0,0 +1,26 @@
+#! /bin/sh
+
+
+if test x"${AUTORECONF}" = x""; then
+    autoreconf -V 1>/dev/null 2>/dev/null
+    if test $? -eq 0; then
+        AUTORECONF=autoreconf
+    fi
+fi
+
+if test x"${AUTORECONF}" != x""; then
+    ${AUTORECONF} -vfi -I m4
+else
+    AUTOCONF=${AUTOCONF:-autoconf}
+    AUTOMAKE=${AUTOMAKE:-automake}
+    AUTOHEADER=${AUTOHEADER:-autoheader}
+    ACLOCAL=${ACLOCAL:-aclocal}
+
+    ${AUTOCONF} --version
+    ${AUTOMAKE} --version
+
+    ${ACLOCAL} -I m4 -I .
+    ${AUTOHEADER} -I .
+    ${AUTOMAKE} --add-missing --copy --force-missing --foreign
+    ${AUTOCONF}
+fi

test/test_text_file.sh

@@ -66,3 +66,6 @@ run_cap_test ${lnav_test} -n \
     -c ";SELECT top_meta FROM lnav_views WHERE name = 'text'" \
     -c ':write-json-to -' \
     < ${test_dir}/example.patch
+
+run_cap_test ${lnav_test} -n \
+    < ${top_srcdir}/autogen.sh