diff options
Diffstat (limited to 'test/test_escape.ml')
| -rw-r--r-- | test/test_escape.ml | 119 |
1 files changed, 119 insertions, 0 deletions
diff --git a/test/test_escape.ml b/test/test_escape.ml new file mode 100644 index 0000000..35a9b9d --- /dev/null +++ b/test/test_escape.ml @@ -0,0 +1,119 @@ +module E = Colitur_render.Escape + +let check = Alcotest.(check string) + +let test_latex () = + check "ampersand" "\\&" (E.apply E.Latex "&"); + check "percent" "\\%" (E.apply E.Latex "%"); + check "underscore" "\\_" (E.apply E.Latex "_"); + check "braces" "\\{\\}" (E.apply E.Latex "{}"); + check "backslash first" "\\textbackslash{}" (E.apply E.Latex "\\"); + (* A real feast name that would otherwise break a .tex build. *) + check "real name" "Ss.mi Nominis Iesu \\& Mari\xc3\xa6" + (E.apply E.Latex "Ss.mi Nominis Iesu & Mari\xc3\xa6") + +let test_groff () = + check "backslash" "\\e" (E.apply E.Groff "\\"); + (* RG-irrelevant but groff-critical: a leading dot starts a request. *) + check "leading dot" "\\&.Ss" (E.apply E.Groff ".Ss"); + check "leading quote" "\\&'tis" (E.apply E.Groff "'tis"); + check "interior dot untouched" "Ss.mi" (E.apply E.Groff "Ss.mi") + +let test_html_xml () = + check "amp first" "&lt;" (E.apply E.Html "<"); + check "angles" "<b>" (E.apply E.Html "<b>"); + check "quote" """ (E.apply E.Html "\""); + check "xml same" "<b>" (E.apply E.Xml "<b>") + +let test_ics () = + check "comma" "\\," (E.apply E.Ics ","); + check "semicolon" "\\;" (E.apply E.Ics ";"); + check "backslash" "\\\\" (E.apply E.Ics "\\"); + check "newline" "\\n" (E.apply E.Ics "\n") + +let test_none_is_identity () = + check "none" "& < > \\ % {}" (E.apply E.None_ "& < > \\ % {}") + +let test_flavour_names () = + List.iter + (fun f -> Alcotest.(check bool) "roundtrip" true (E.of_string (E.to_string f) = Some f)) + [ E.Latex; E.Groff; E.Html; E.Xml; E.Ics; E.None_ ]; + Alcotest.(check bool) "tex" true (E.of_extension ".tex" = Some E.Latex); + Alcotest.(check bool) "ms" true (E.of_extension ".ms" = Some E.Groff); + Alcotest.(check bool) "mom" true (E.of_extension ".mom" = Some E.Groff); + Alcotest.(check bool) "html" true (E.of_extension ".html" = Some E.Html); + Alcotest.(check bool) "md is none" true (E.of_extension ".md" = Some E.None_); + Alcotest.(check bool) "adoc is none" true (E.of_extension ".adoc" = Some E.None_); + Alcotest.(check bool) "txt is none" true (E.of_extension ".txt" = Some E.None_); + (* Spec section 5: an unknown extension is an ERROR, never a silent fallback. *) + Alcotest.(check bool) "unknown is None" true (E.of_extension ".wat" = None) + +let test_fold_short_line_unchanged () = + check "short" "SUMMARY:Feast\r\n" (E.fold_ics "SUMMARY:Feast") + +let test_fold_long_line () = + let long = "DESCRIPTION:" ^ String.make 200 'x' in + let out = E.fold_ics long in + let lines = String.split_on_char '\n' out in + List.iter + (fun l -> + let l = if l <> "" && l.[String.length l - 1] = '\r' then String.sub l 0 (String.length l - 1) else l in + if String.length l > 75 then Alcotest.failf "line of %d octets exceeds 75" (String.length l)) + (List.filter (fun l -> l <> "") lines); + (* Continuation lines begin with exactly one space (RFC 5545 section 3.1). *) + match lines with + | _ :: second :: _ -> Alcotest.(check bool) "continuation starts with space" true (second.[0] = ' ') + | _ -> Alcotest.fail "expected the line to fold" + +(* RFC 5545 section 3.1 folds on OCTET boundaries; splitting mid-UTF-8 corrupts + Polish and Latin names, which is the whole reason this is not a template job. *) +let test_fold_never_splits_utf8 () = + let long = "SUMMARY:" ^ String.concat "" (List.init 40 (fun _ -> "\xc4\x99\xc5\x9b\xc4\x87")) in + let out = E.fold_ics long in + let chunks = + List.filter_map + (fun l -> + let l = if l <> "" && l.[String.length l - 1] = '\r' then String.sub l 0 (String.length l - 1) else l in + if l = "" then None else if l.[0] = ' ' then Some (String.sub l 1 (String.length l - 1)) else Some l) + (String.split_on_char '\n' out) + in + (* The named property: no chunk may START with a UTF-8 continuation byte + (0x80-0xBF) -- that would mean the previous fold cut mid-character. + Unfolding losslessly (below) cannot detect this on its own: concatenation + is insensitive to where the cuts fell, so a fold at ANY position still + round-trips. *) + List.iter + (fun c -> + if String.length c > 0 then + Alcotest.(check bool) "chunk does not start mid-UTF-8" true (Char.code c.[0] land 0xC0 <> 0x80)) + chunks; + let stripped = String.concat "" chunks in + check "unfolds to the original" long stripped + +(* B1 regression: 100 consecutive UTF-8 continuation bytes (0x80-0xBF) is not + producible by valid UTF-8 (whose longest continuation run is 3), but + fold_ics must stay TOTAL on arbitrary octet strings. A backoff loop with no + hard-cut fallback backs `cut` all the way down to `pos`, yielding a + zero-length chunk and recursing on the identical position forever. *) +let test_fold_pathological_input_terminates () = + let pathological = String.make 100 '\x80' in + let out = E.fold_ics pathological in + let lines = String.split_on_char '\n' out in + List.iter + (fun l -> + let l = if l <> "" && l.[String.length l - 1] = '\r' then String.sub l 0 (String.length l - 1) else l in + if String.length l > 75 then Alcotest.failf "line of %d octets exceeds 75" (String.length l)) + (List.filter (fun l -> l <> "") lines) + +let suite = + ( "Escape", + [ Alcotest.test_case "latex" `Quick test_latex; + Alcotest.test_case "groff" `Quick test_groff; + Alcotest.test_case "html/xml" `Quick test_html_xml; + Alcotest.test_case "ics" `Quick test_ics; + Alcotest.test_case "none is identity" `Quick test_none_is_identity; + Alcotest.test_case "flavour names" `Quick test_flavour_names; + Alcotest.test_case "fold: short unchanged" `Quick test_fold_short_line_unchanged; + Alcotest.test_case "fold: long line" `Quick test_fold_long_line; + Alcotest.test_case "fold: never splits utf8" `Quick test_fold_never_splits_utf8; + Alcotest.test_case "fold: pathological input terminates" `Quick test_fold_pathological_input_terminates ] ) |