| 1 | // Package export turns a document's structure into the formats a |
| 2 | // person can take away with them: Markdown today, PDF alongside it |
| 3 | // (pdf.go), DOCX when §11.2's import work makes a round trip worth |
| 4 | // having. |
| 5 | // |
| 6 | // The structure arrives as ProseMirror-shaped JSON from the Rust core |
| 7 | // (core/src/fragment.rs's fragment_json, read through |
| 8 | // internal/core's Handle.JSON) — not as the HTML the read view |
| 9 | // renders, because "where does this list end" and "is this run bold" |
| 10 | // are questions the markup answers only to a parser, and Go's standard |
| 11 | // library has none. |
| 12 | package export |
| 13 | |
| 14 | import ( |
| 15 | "encoding/json" |
| 16 | "fmt" |
| 17 | ) |
| 18 | |
| 19 | // Node is one node of the document tree: an element with children, or |
| 20 | // a text run with marks. It is the JSON the core emits, decoded — a |
| 21 | // deliberately loose mirror rather than a validated schema, because |
| 22 | // the core carries unknown nodes and unknown marks through on purpose |
| 23 | // (a newer editor build must not lose text through an older |
| 24 | // exporter), and a decoder that rejected them would undo that. |
| 25 | type Node struct { |
| 26 | Type string `json:"type"` |
| 27 | Attrs Attrs `json:"attrs"` |
| 28 | Content []Node `json:"content"` |
| 29 | Text string `json:"text"` |
| 30 | Marks []Mark `json:"marks"` |
| 31 | } |
| 32 | |
| 33 | // Attrs is the small, closed set of node attributes the core emits. |
| 34 | // |
| 35 | // Indent is a level, not a measurement: the core emits 1–8 and both |
| 36 | // renderers turn that into half an inch apiece. Keeping it a level here |
| 37 | // rather than a width in points is what lets the screen and the PDF |
| 38 | // disagree about the unit without disagreeing about the document — |
| 39 | // views/show.html works in inches and this exporter in points, and |
| 40 | // neither has to know the other's. |
| 41 | type Attrs struct { |
| 42 | Level int `json:"level"` |
| 43 | Align string `json:"align"` |
| 44 | Indent int `json:"indent"` |
| 45 | } |
| 46 | |
| 47 | // Mark is one mark on a text run. Only the name matters — every mark |
| 48 | // this build knows is attribute-free by design; see the MARKS table in |
| 49 | // core/src/fragment.rs for why. |
| 50 | type Mark struct { |
| 51 | Type string `json:"type"` |
| 52 | } |
| 53 | |
| 54 | // Has reports whether a run carries a named mark. |
| 55 | func (n Node) Has(mark string) bool { |
| 56 | for _, m := range n.Marks { |
| 57 | if m.Type == mark { |
| 58 | return true |
| 59 | } |
| 60 | } |
| 61 | return false |
| 62 | } |
| 63 | |
| 64 | // IsText reports whether this node is a text run rather than an |
| 65 | // element. Checking the type rather than "does it have children" — |
| 66 | // an empty paragraph has no children either, and is not text. |
| 67 | func (n Node) IsText() bool { return n.Type == "text" } |
| 68 | |
| 69 | // Document is a whole document: its title (from the database row, not |
| 70 | // from the CRDT — the title is not part of the content) and the root |
| 71 | // node the core rendered. |
| 72 | type Document struct { |
| 73 | Title string |
| 74 | Root Node |
| 75 | } |
| 76 | |
| 77 | // Parse decodes the core's JSON into a Document. |
| 78 | func Parse(title, structureJSON string) (Document, error) { |
| 79 | var root Node |
| 80 | if err := json.Unmarshal([]byte(structureJSON), &root); err != nil { |
| 81 | return Document{}, fmt.Errorf("export: decode document structure: %w", err) |
| 82 | } |
| 83 | if root.Type != "doc" { |
| 84 | return Document{}, fmt.Errorf("export: document structure has root type %q, want \"doc\"", root.Type) |
| 85 | } |
| 86 | return Document{Title: title, Root: root}, nil |
| 87 | } |
| 88 | |
| 89 | // blockKind classifies a node for the writers below. Both Markdown and |
| 90 | // PDF ask the same questions of the tree, and asking them in one place |
| 91 | // is what keeps the two exports describing the same document. |
| 92 | type blockKind int |
| 93 | |
| 94 | const ( |
| 95 | blockUnknown blockKind = iota |
| 96 | blockParagraph |
| 97 | blockHeading |
| 98 | blockBulletList |
| 99 | blockOrderedList |
| 100 | blockListItem |
| 101 | blockQuote |
| 102 | ) |
| 103 | |
| 104 | func kindOf(n Node) blockKind { |
| 105 | switch n.Type { |
| 106 | case "paragraph": |
| 107 | return blockParagraph |
| 108 | case "heading": |
| 109 | return blockHeading |
| 110 | case "bullet_list": |
| 111 | return blockBulletList |
| 112 | case "ordered_list": |
| 113 | return blockOrderedList |
| 114 | case "list_item": |
| 115 | return blockListItem |
| 116 | case "blockquote": |
| 117 | return blockQuote |
| 118 | } |
| 119 | return blockUnknown |
| 120 | } |
| 121 | |
| 122 | // headingLevel clamps to 1–6 a second time. The core already clamps |
| 123 | // (fragment.rs), so this can only fire on a level the core did not |
| 124 | // produce — but the two are separate binaries' worth of trust apart |
| 125 | // once the JSON has crossed the ABI, and a "#######" in a Markdown |
| 126 | // file is not a heading at all. |
| 127 | // The two alignments that are not the default. core/src/fragment.rs's |
| 128 | // ALIGNMENTS table is where this vocabulary is defined; these are the |
| 129 | // same two values, spelled the same way, because they arrive from it |
| 130 | // over the wire. |
| 131 | const ( |
| 132 | alignCentre = "center" |
| 133 | alignRight = "right" |
| 134 | ) |
| 135 | |
| 136 | // alignOf is the allowlist a second time, on this side of the ABI, and |
| 137 | // it is exact — no case folding, no trimming, no synonyms. |
| 138 | // |
| 139 | // It exists for the same reason headingLevel clamps a level the core |
| 140 | // has already clamped: by the time the JSON reaches this package it has |
| 141 | // crossed a binary boundary, and this package's own decoder is |
| 142 | // deliberately loose (Node carries unknown types and unknown marks |
| 143 | // through, so a newer editor cannot lose a reader's text). Loose about |
| 144 | // what it CARRIES, strict about what it ACTS on: an alignment this |
| 145 | // build does not know lays out as left rather than as a guess, and |
| 146 | // "CENTER" is not "center" — a normalisation here would be this |
| 147 | // package quietly deciding a question the schema and the core have both |
| 148 | // already answered, and would disagree with the read view, which drops |
| 149 | // it. A document must not paginate one way on screen and another in its |
| 150 | // own export. |
| 151 | // |
| 152 | // Empty means left, which is both "no attribute" and "align=left" — |
| 153 | // the default spends nothing, exactly as it spends no class in the read |
| 154 | // view. |
| 155 | func alignOf(n Node) string { |
| 156 | switch n.Attrs.Align { |
| 157 | case alignCentre: |
| 158 | return alignCentre |
| 159 | case alignRight: |
| 160 | return alignRight |
| 161 | } |
| 162 | return "" |
| 163 | } |
| 164 | |
| 165 | func headingLevel(n Node) int { |
| 166 | l := n.Attrs.Level |
| 167 | if l < 1 { |
| 168 | return 1 |
| 169 | } |
| 170 | if l > 6 { |
| 171 | return 6 |
| 172 | } |
| 173 | return l |
| 174 | } |
| 175 | |
amadan