xref: /freebsd/contrib/lyaml/spec/ext_yaml_parser_spec.yaml (revision 2bc180ef045e5911cce0cea1c2a139cffd2b577a)

# LYAML binding for Lua 5.1, 5.2, 5.3 & 5.4
# Copyright (C) 2013-2022 Gary V. Vaughan

specify parsing:
- it parses empty streams:
    e = yaml.parser ""
    expect (e ().type).to_be "STREAM_START"
    expect (e ().type).to_be "STREAM_END"
    expect (e ()).to_be (nil)
    expect (e ()).to_be (nil)
- it ignores comments: '
    e = yaml.parser "# A comment\nnon-comment # trailing comment\n"
    expect (e ().type).to_be "STREAM_START"
    expect (e ().type).to_be "DOCUMENT_START"
    expect (e ().value).to_be "non-comment"
    expect (e ().type).to_be "DOCUMENT_END"'

- describe STREAM_START:
  - before:
      e = yaml.parser "# no BOM"
  - it is the first event:
      expect (e ().type).to_be "STREAM_START"
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 0, column = 0, index = 0}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 0, column = 0, index = 0}
  - it uses UTF-8 by default:
      expect (e ().encoding).to_be "UTF8"
  - it recognizes UTF-16 BOM:
      e = yaml.parser (BOM .. " BOM")
      expect (e ().encoding).to_match "UTF16[BL]E"

- describe STREAM_END:
  - before:
      for t in yaml.parser "nothing to see" do ev = t end
  - it is the last event:
      expect (ev.type).to_be "STREAM_END"
  - it reports event start marker:
      expect (ev.start_mark).to_equal {line = 1, column = 0, index = 14}
  - it reports event end marker:
      expect (ev.end_mark).to_equal {line = 1, column = 0, index = 14}

- describe DOCUMENT_START:
  - before:
      e = consume (1, "---")
  - it recognizes document start marker:
      expect (filter (e (), "type", "implicit")).
        to_equal {type = "DOCUMENT_START", implicit = false}
  - it reports implicit document start:
      e = consume (1, "foo")
      expect (e ().implicit).to_be (true)
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 0, column = 0, index = 0}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 0, column = 3, index = 3}

  - context parser directives:
    - it can recognize document versions:
        e = consume (1, "%YAML 1.1\n---")
        expect (e ().version_directive).to_equal {major = 1, minor = 1}
    - it can diagnose missing document start:
        e = consume (1, "%YAML 1.1\n")
        expect (e ()).to_error "expected <document start>"
    - it can diagnose multiple versions:
        e = consume (1, "%YAML 1.1\n%YAML 1.1\n---")
        expect (e ()).to_error "duplicate %YAML directive"
    - it can diagnose too-new versions:
        e = consume (1, "%YAML 2.0\n---")
        expect (e ()).to_error "incompatible YAML document"
    - it warns of newer minor versions:
        pending (github_issue "1")
        e = consume (1, "%YAML 1.9\n---")
        expect (e ()).
           to_error "attempting parsing of newer minor document version"

    - it can recognize primary tag handles:
        e = consume (1, "%TAG ! tag:ben-kiki.org,2000:app/\n---")
        expect (e ().tag_directives).
           to_equal {{handle = "!", prefix = "tag:ben-kiki.org,2000:app/"}}
    - it can recognize secondary tag handles:
        e = consume (1, "%TAG !! tag:yaml.org,2002:\n---")
        expect (e ().tag_directives).
           to_equal {{handle = "!!", prefix = "tag:yaml.org,2002:"}}
    - it can recognize named tag handles:
        e = consume (1, "%TAG !o! tag:ben-kiki.org,2000:\n---")
        expect (e ().tag_directives).
           to_equal {{handle = "!o!", prefix = "tag:ben-kiki.org,2000:"}}
    - it can concatenate multiple tag handles:
        e = consume (1, "%TAG ! !\n" ..
                        "%TAG !! tag:yaml.org,2002:\n" ..
                        "%TAG !o! tag:ben-kiki.org,2000:\n" ..
                        "---")
        expect (e ().tag_directives).to_contain.
           all_of {{handle = "!", prefix = "!"},
                   {handle = "!!", prefix = "tag:yaml.org,2002:"},
                   {handle = "!o!", prefix = "tag:ben-kiki.org,2000:"}}
    - it can diagnose missing document start:
        e = consume (1, "%TAG ! !\n")
        expect (e ()).to_error "expected <document start>"

- describe DOCUMENT_END:
  - before:
      e = consume (3, "foo\n...")
  - it recognizes the document end marker:
      expect (filter (e (), "type", "implicit")).
         to_equal {type = "DOCUMENT_END", implicit = false}
  - it reports an implicit document end marker:
      e = consume (3, "foo\n")
      expect (filter (e (), "type", "implicit")).
         to_equal {type = "DOCUMENT_END", implicit = true}
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 1, column = 0, index = 4}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 1, column = 3, index = 7}

- describe ALIAS:
  - before:
      e = consume (10, "---\n" ..
                      "hr:\n" ..
                      "- Mark McGwire\n" ..
                      "- &SS Sammy Sosa\n" ..
                      "rbi:\n" ..
                      "- *SS\n" ..
                      "- Ken Griffey")
  - it recognizes an alias event:
      expect (filter (e (), "type", "anchor")).
         to_equal {type = "ALIAS", anchor = "SS"}
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 5, column = 2, index = 47}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 5, column = 5, index = 50}

- describe SCALAR:
  - before:
      e = consume (6, "---\n" ..
                      "hr:\n" ..
                      "- Mark McGwire\n" ..
                      "- &SS Sammy Sosa\n" ..
                      "rbi:\n" ..
                      "- *SS\n" ..
                      "- Ken Griffey")
  - it recognizes a scalar event:
      expect (filter (e (), "type", "value")).
         to_equal {type = "SCALAR",  value = "Sammy Sosa"}
  - it records anchors:
      expect (e ().anchor).to_be "SS"
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 3, column = 2, index = 25}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 3, column = 16, index = 39}

  - context with quoting style:
    - context plain style:
      - before:
          e = consume (2, "---\n" ..
                          "  Mark McGwire's\n" ..
                          "  year was crippled\n" ..
                          "  by a knee injury.\n")
      - it ignores line-breaks and indentation:
          expect (e ().value).
             to_be "Mark McGwire's year was crippled by a knee injury."
      - it recognizes implicit plain style:
          e = consume (2, "---\n" ..
                          "  Mark McGwire's\n" ..
                          "  year was crippled\n" ..
                          "  by a knee injury.\n")
          expect (e ().plain_implicit).to_be (true)
      - it recognizes explicit plain style:
          e = consume (2, "|\n" ..
                          "  Mark McGwire's\n" ..
                          "  year was crippled\n" ..
                          "  by a knee injury.\n")
          expect (e ().plain_implicit).to_be (false)
      - it recognizes implicit quoted style:
          e = consume (2, "|\n" ..
                          "  Mark McGwire's\n" ..
                          "  year was crippled\n" ..
                          "  by a knee injury.\n")
          expect (e ().quoted_implicit).to_be (true)
      - it recognizes explicit quoted style:
          e = consume (2, "'\n" ..
                          "  Mark McGwire's\n" ..
                          "  year was crippled\n" ..
                          "  by a knee injury.'\n")
          expect (e ().plain_implicit).to_be (false)
    - context folded style:
      - it preserves blank lines and deeper indentation:
          e = consume (2, ">\n" ..
                          "  Sammy Sosa completed another\n" ..
                          "  fine season with great stats.\n" ..
                          "\n" ..
                          "    63 Home Runs\n" ..
                          "    0.288 Batting Average\n" ..
                          "\n" ..
                          "  What a year!\n")
          expect (e ().value).
             to_be ("Sammy Sosa completed another fine season with great stats.\n" ..
                       "\n" ..
                       "  63 Home Runs\n" ..
                       "  0.288 Batting Average\n" ..
                       "\n" ..
                       "What a year!\n")
    - context literal style:
      - it removes indentation but preserves all line-breaks:
          e = consume (2, [[# ASCII Art]] .. "\n" ..
                          [[--- |]] .. "\n" ..
                          [[  \//||\/||]] .. "\n" ..
                          [[  // ||  ||__]] .. "\n")
          expect (e ().value).
             to_be ([[\//||\/||]] .. "\n" ..
                        [[// ||  ||__]] .. "\n")

    - context single quoted style:
      - it folds line breaks:
          e = consume (2, [['This quoted scalar]] .. "\n" ..
                          [[  spans two lines.']])
          expect (e ().value).
             to_be "This quoted scalar spans two lines."
      - it does not process escape sequences:
          # Lua [[ quoting makes sure libyaml sees all the quotes.
          e = consume (2, [['"Howdy!"\t\u263A']])
          expect (e ().value).to_be [["Howdy!"\t\u263A]]

    # Note that we have to single quote the Lua snippets to prevent
    # libyaml from interpreting the bytes as the spec file is read, so
    # that the raw strings get correctly passed to the Lua compiler.
    - context double quoted style:
      - it folds line breaks: '
          e = consume (4, [[quoted: "This quoted scalar]] .. "\n" ..
                          [[  spans two lines\n"]])
          expect (e ().value).
             to_be "This quoted scalar spans two lines\n"'
      - it recognizes unicode escape sequences: '
          e = consume (4, [[unicode: "Sosa did fine.\u263A"]])
          expect (e ().value).to_be "Sosa did fine.\226\152\186"'
      - it recognizes control escape sequences: '
          e = consume (4, [[control: "\b1998\t1999\t2000\n"]])
          expect (e ().value).to_be "\b1998\t1999\t2000\n"'
      - it recognizes hexadecimal escape sequences: '
          e = consume (4, [[hexesc: "\x41\x42\x43 is ABC"]])
          expect (e ().value).to_be "ABC is ABC"'

    - context indentation determines scope: '
        e = consume (4, "name: Mark McGwire\n" ..
                        "accomplishment: >\n" ..
                        "  Mark set a major league\n" ..
                        "  home run record in 1998.\n" ..
                        "stats: |\n" ..
                        "  65 Home Runs\n" ..
                        "  0.278 Batting Average\n")
        expect (e ().value).to_be "Mark McGwire"
        expect (e ().value).to_be "accomplishment"
        expect (e ().value).
           to_be "Mark set a major league home run record in 1998.\n"
        expect (e ().value).to_be "stats"
        expect (e ().value).to_be "65 Home Runs\n0.278 Batting Average\n"'

  - context with tag:
    - it recognizes local tags: '
        e = consume (4, "application specific tag: !something |\n" ..
                        " The semantics of the tag\n" ..
                        " above may be different for\n" ..
                        " different documents.")
        expect (e ().tag).to_be "!something"'
    - it recognizes global tags: '
        e = consume (4, "picture: !!binary |\n" ..
                        " R0lGODlhDAAMAIQAAP//9/X\n" ..
                        " 17unp5WZmZgAAAOfn515eXv\n" ..
                        " Pz7Y6OjuDg4J+fn5OTk6enp\n" ..
                        " 56enmleECcgggoBADs=")
        expect (e ().tag).to_be "tag:yaml.org,2002:binary"'
    - it resolves %TAG declarations: '
        e = consume (5, "%TAG ! tag:clarkevans.com,2002:\n" ..
                        "---\n" ..
                        "shape:\n" ..
                        "- !circle\n" ..
                        "  center: &ORIGIN {x: 73, y: 129}\n" ..
                        "  radius: 7")
        expect (e ().tag).to_be "tag:clarkevans.com,2002:circle"'

- describe SEQUENCE_START:
  - before: '
      e = consume (4, "fubar: &FOO\n" ..
                      "  - foo\n" ..
                      "  - bar\n")'
  - it recognizes a sequence start event:
      expect (e ().type).to_be "SEQUENCE_START"
  - it records anchors:
      expect (e ().anchor).to_be "FOO"
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 0, column = 7, index = 7}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 1, column = 2, index = 14}

  - context with tag:
    - it recognizes local tags: '
        e = consume (2, "--- !something\n" ..
                        "- foo\n")
        expect (filter (e (), "type", "tag")).
           to_equal {type = "SEQUENCE_START", tag = "!something"}'
    - it recognizes global tags: '
        e = consume (2, "--- !!omap\n" ..
                        "- Mark McGwire: 65\n" ..
                        "- Sammy Sosa: 63\n" ..
                        "- Ken Griffy: 58\n")
        expect (filter (e (), "type", "tag")).
           to_equal {type = "SEQUENCE_START",
                        tag  = "tag:yaml.org,2002:omap"}'
    - it resolves %TAG declarations: '
        e = consume (2, "%TAG ! tag:clarkevans.com,2002:\n" ..
                        "--- !shape\n" ..
                        "- !circle\n" ..
                        "  center: &ORIGIN {x: 73, y: 129}\n" ..
                        "  radius: 7\n")
        expect (filter (e (), "type", "tag")).
           to_equal {type = "SEQUENCE_START",
                        tag  = "tag:clarkevans.com,2002:shape"}'

  - context with style:
    - it recognizes block style:
        e = consume (2, "- first\n- second")
        expect (filter (e (), "type", "style")).
           to_equal {type = "SEQUENCE_START", style = "BLOCK"}
    - it recognizes flow style:
        e = consume (2, "[first, second]")
        expect (filter (e (), "type", "style")).
           to_equal {type = "SEQUENCE_START", style = "FLOW"}

- describe SEQUENCE_END:
  - before:
      e = consume (5, "- foo\n- bar\n")
  - it recognizes a sequence end event:
      expect (e ().type).to_equal "SEQUENCE_END"
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 2, column = 0, index = 12}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 2, column = 0, index = 12}

- describe MAPPING_START:
  - before: 'e = consume (3, "- &FUBAR\n  foo: bar\n")'
  - it recognizes a mapping start event:
        expect (e ().type).to_be "MAPPING_START"
  - it records anchors:
      expect (e ().anchor).to_be "FUBAR"
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 0, column = 2, index = 2}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 1, column = 2, index = 11}

  - context with tag:
    - it recognizes local tags: '
    e = consume (2, "--- !something\nfoo: bar\n")
        expect (filter (e (), "type", "tag")).
           to_equal {type = "MAPPING_START", tag = "!something"}'
    - it recognizes global tags: '
        e = consume (2, "--- !!set\n" ..
                        "? Mark McGwire\n" ..
                        "? Sammy Sosa\n" ..
                        "? Ken Griffy\n")
        expect (filter (e (), "type", "tag")).
           to_equal {type = "MAPPING_START",
                        tag  = "tag:yaml.org,2002:set"}'
    - it resolves %TAG declarations: '
        e = consume (3, "%TAG ! tag:clarkevans.com,2002:\n" ..
                        "--- !shape\n" ..
                        "- !circle\n" ..
                        "  center: &ORIGIN {x: 73, y: 129}\n" ..
                        "  radius: 7\n")
        expect (filter (e (), "type", "tag")).
           to_equal {type = "MAPPING_START",
                        tag  = "tag:clarkevans.com,2002:circle"}'

  - context with style:
    - it recognizes block style: '
        e = consume (2, "foo: bar\nbaz:\n  quux")
        expect (filter (e (), "type", "style")).
           to_equal {type = "MAPPING_START", style = "BLOCK"}'
    - it recognizes flow style: '
        e = consume (2, "{foo: bar, baz: quux}")
        expect (filter (e (), "type", "style")).
           to_equal {type = "MAPPING_START", style = "FLOW"}'


- describe MAPPING_END:
  - before: 'e = consume (5, "foo: bar\n")'
  - it recognizes the mapping end event:
      expect (e ().type).to_equal "MAPPING_END"
  - it reports event start marker:
      expect (e ().start_mark).to_equal {line = 1, column = 0, index = 9}
  - it reports event end marker:
      expect (e ().end_mark).to_equal {line = 1, column = 0, index = 9}