From 1b5ee524dd76729a4d790e14245a5a9d7b6dd974 Mon Sep 17 00:00:00 2001 From: Michael Williamson Date: Mon, 28 Mar 2016 23:20:56 +0100 Subject: Move conversion classes to conversion namespace --- .../org/zwobble/mammoth/DocumentConverter.java | 4 +- .../zwobble/mammoth/internal/DocumentToHtml.java | 249 --------------------- .../mammoth/internal/DocumentToHtmlOptions.java | 53 ----- .../internal/conversion/DocumentToHtml.java | 249 +++++++++++++++++++++ .../internal/conversion/DocumentToHtmlOptions.java | 53 +++++ 5 files changed, 304 insertions(+), 304 deletions(-) delete mode 100644 src/main/java/org/zwobble/mammoth/internal/DocumentToHtml.java delete mode 100644 src/main/java/org/zwobble/mammoth/internal/DocumentToHtmlOptions.java create mode 100644 src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java create mode 100644 src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtmlOptions.java (limited to 'src/main/java/org') diff --git a/src/main/java/org/zwobble/mammoth/DocumentConverter.java b/src/main/java/org/zwobble/mammoth/DocumentConverter.java index 46ec83c..a357487 100644 --- a/src/main/java/org/zwobble/mammoth/DocumentConverter.java +++ b/src/main/java/org/zwobble/mammoth/DocumentConverter.java @@ -1,7 +1,7 @@ package org.zwobble.mammoth; -import org.zwobble.mammoth.internal.DocumentToHtml; -import org.zwobble.mammoth.internal.DocumentToHtmlOptions; +import org.zwobble.mammoth.internal.conversion.DocumentToHtml; +import org.zwobble.mammoth.internal.conversion.DocumentToHtmlOptions; import org.zwobble.mammoth.internal.documents.DocumentElement; import org.zwobble.mammoth.internal.documents.HasChildren; import org.zwobble.mammoth.internal.documents.Paragraph; diff --git a/src/main/java/org/zwobble/mammoth/internal/DocumentToHtml.java b/src/main/java/org/zwobble/mammoth/internal/DocumentToHtml.java deleted file mode 100644 index 27d9ccc..0000000 --- a/src/main/java/org/zwobble/mammoth/internal/DocumentToHtml.java +++ /dev/null @@ -1,249 +0,0 @@ -package org.zwobble.mammoth.internal; - -import org.zwobble.mammoth.internal.documents.*; -import org.zwobble.mammoth.internal.html.Html; -import org.zwobble.mammoth.internal.html.HtmlNode; -import org.zwobble.mammoth.internal.results.InternalResult; -import org.zwobble.mammoth.internal.styles.HtmlPath; -import org.zwobble.mammoth.internal.styles.StyleMap; -import org.zwobble.mammoth.internal.util.Lists; - -import java.io.IOException; -import java.util.*; - -import static org.zwobble.mammoth.internal.util.Lists.*; -import static org.zwobble.mammoth.internal.util.Maps.map; -import static org.zwobble.mammoth.internal.util.Streams.toByteArray; - -public class DocumentToHtml { - public static InternalResult> convertToHtml(Document document, DocumentToHtmlOptions options) { - DocumentToHtml documentConverter = new DocumentToHtml(options); - return new InternalResult<>( - documentConverter.convertToHtml(document), - documentConverter.warnings); - } - - private static List findNotes(Document document, Iterable noteReferences) { - return eagerMap( - noteReferences, - // TODO: handle missing notes - reference -> document.getNotes().findNote(reference.getNoteType(), reference.getNoteId()).get()); - } - - public static InternalResult> convertToHtml(DocumentElement element, DocumentToHtmlOptions options) { - DocumentToHtml documentConverter = new DocumentToHtml(options); - return new InternalResult<>( - documentConverter.convertToHtml(element), - documentConverter.warnings); - } - - private final String idPrefix; - private final boolean preserveEmptyParagraphs; - private final StyleMap styleMap; - private final List noteReferences = new ArrayList<>(); - private final Set warnings = new HashSet<>(); - - private DocumentToHtml(DocumentToHtmlOptions options) { - this.idPrefix = options.idPrefix(); - this.preserveEmptyParagraphs = options.shouldPreserveEmptyParagraphs(); - this.styleMap = options.styleMap(); - } - - private List convertToHtml(Document document) { - List mainBody = convertChildrenToHtml(document); - // TODO: can you have note references inside a note? - List notes = findNotes(document, noteReferences); - if (notes.isEmpty()) { - return mainBody; - } else { - HtmlNode noteNode = Html.element("ol", - eagerMap(notes, this::convertToHtml)); - - return eagerConcat(mainBody, list(noteNode)); - } - } - - private HtmlNode convertToHtml(Note note) { - String id = generateNoteHtmlId(note.getNoteType(), note.getId()); - String referenceId = generateNoteRefHtmlId(note.getNoteType(), note.getId()); - List noteBody = convertToHtml(note.getBody()); - // TODO: we probably want this to collapse more eagerly than other collapsible elements - // -- for instance, any paragraph will probably do, regardless of attributes. (Possible other elements will do too.) - HtmlNode backLink = Html.collapsibleElement("p", list( - Html.text(" "), - Html.element("a", map("href", "#" + referenceId), list(Html.text("↑"))))); - return Html.element("li", map("id", id), eagerConcat(noteBody, list(backLink))); - } - - private List convertToHtml(List elements) { - return eagerFlatMap( - elements, - this::convertToHtml); - } - - private List convertChildrenToHtml(HasChildren element) { - return convertToHtml(element.getChildren()); - } - - private List convertToHtml(DocumentElement element) { - return element.accept(new DocumentElementVisitor>() { - @Override - public List visit(Paragraph paragraph) { - List content = convertChildrenToHtml(paragraph); - List children = preserveEmptyParagraphs ? cons(Html.FORCE_WRITE, content) : content; - HtmlPath mapping = styleMap.getParagraphHtmlPath(paragraph) - .orElseGet(() -> { - if (paragraph.getStyle().isPresent()) { - warnings.add("Unrecognised paragraph style: " + paragraph.getStyle().get().describe()); - } - return HtmlPath.element("p"); - }); - return mapping.wrap(children); - } - - @Override - public List visit(Run run) { - List nodes = convertChildrenToHtml(run); - if (run.isStrikethrough()) { - nodes = styleMap.getStrikethrough().orElse(HtmlPath.collapsibleElement("s")).wrap(nodes); - } - if (run.isUnderline()) { - nodes = styleMap.getUnderline().orElse(HtmlPath.EMPTY).wrap(nodes); - } - if (run.getVerticalAlignment() == VerticalAlignment.SUBSCRIPT) { - nodes = list(Html.collapsibleElement("sub", nodes)); - } - if (run.getVerticalAlignment() == VerticalAlignment.SUPERSCRIPT) { - nodes = list(Html.collapsibleElement("sup", nodes)); - } - if (run.isItalic()) { - nodes = styleMap.getItalic().orElse(HtmlPath.collapsibleElement("em")).wrap(nodes); - } - if (run.isBold()) { - nodes = styleMap.getBold().orElse(HtmlPath.collapsibleElement("strong")).wrap(nodes); - } - HtmlPath mapping = styleMap.getRunHtmlPath(run) - .orElseGet(() -> { - if (run.getStyle().isPresent()) { - warnings.add("Unrecognised run style: " + run.getStyle().get().describe()); - } - return HtmlPath.EMPTY; - }); - return mapping.wrap(nodes); - } - - @Override - public List visit(Text text) { - if (text.getValue().isEmpty()) { - return list(); - } else { - return list(Html.text(text.getValue())); - } - } - - @Override - public List visit(Tab tab) { - return list(Html.text("\t")); - } - - @Override - public List visit(LineBreak lineBreak) { - return list(Html.selfClosingElement("br")); - } - - @Override - public List visit(Table table) { - return list(Html.element("table", convertChildrenToHtml(table))); - } - - @Override - public List visit(TableRow tableRow) { - return list(Html.element("tr", convertChildrenToHtml(tableRow))); - } - - @Override - public List visit(TableCell tableCell) { - return list(Html.element("td", - Lists.cons(Html.FORCE_WRITE, convertChildrenToHtml(tableCell)))); - } - - @Override - public List visit(Hyperlink hyperlink) { - Map attributes = map("href", generateHref(hyperlink)); - return list(Html.collapsibleElement("a", attributes, convertChildrenToHtml(hyperlink))); - } - - private String generateHref(Hyperlink hyperlink) { - if (hyperlink.getHref().isPresent()) { - return hyperlink.getHref().get(); - } else if (hyperlink.getAnchor().isPresent()) { - return "#" + generateId(hyperlink.getAnchor().get()); - } else { - return ""; - } - } - - @Override - public List visit(Bookmark bookmark) { - return list(Html.element("a", map("id", generateId(bookmark.getName())), list(Html.FORCE_WRITE))); - } - - @Override - public List visit(NoteReference noteReference) { - noteReferences.add(noteReference); - String noteAnchor = generateNoteHtmlId(noteReference.getNoteType(), noteReference.getNoteId()); - String noteReferenceAnchor = generateNoteRefHtmlId(noteReference.getNoteType(), noteReference.getNoteId()); - return list(Html.element("sup", list( - Html.element("a", map("href", "#" + noteAnchor, "id", noteReferenceAnchor), list( - Html.text("[" + noteReferences.size() + "]")))))); - } - - @Override - public List visit(Image image) { - // TODO: custom image handlers - // TODO: handle empty content type - return image.getContentType() - .map(contentType -> { - try { - Map attributes = new HashMap<>(); - - String base64 = Base64.getEncoder().encodeToString(toByteArray(image.open())); - String src = "data:" + contentType + ";base64," + base64; - attributes.put("src", src); - - image.getAltText().ifPresent(altText -> attributes.put("alt", altText)); - - return list(Html.selfClosingElement("img", attributes)); - } catch (IOException exception) { - warnings.add(exception.getMessage()); - return Lists.list(); - } - }) - .orElse(list()); - } - }); - } - - private String generateNoteHtmlId(NoteType noteType, String noteId) { - return generateId(noteTypeToIdFragment(noteType) + "-" + noteId); - } - - private String generateNoteRefHtmlId(NoteType noteType, String noteId) { - return generateId(noteTypeToIdFragment(noteType) + "-ref-" + noteId); - } - - private String noteTypeToIdFragment(NoteType noteType) { - switch (noteType) { - case FOOTNOTE: - return "footnote"; - case ENDNOTE: - return "endnote"; - default: - throw new UnsupportedOperationException(); - } - } - - private String generateId(String bookmarkName) { - return idPrefix + bookmarkName; - } -} diff --git a/src/main/java/org/zwobble/mammoth/internal/DocumentToHtmlOptions.java b/src/main/java/org/zwobble/mammoth/internal/DocumentToHtmlOptions.java deleted file mode 100644 index 05218a9..0000000 --- a/src/main/java/org/zwobble/mammoth/internal/DocumentToHtmlOptions.java +++ /dev/null @@ -1,53 +0,0 @@ -package org.zwobble.mammoth.internal; - -import org.zwobble.mammoth.internal.styles.DefaultStyles; -import org.zwobble.mammoth.internal.styles.StyleMap; -import org.zwobble.mammoth.internal.styles.parsing.StyleMapParser; - -public class DocumentToHtmlOptions { - public static final DocumentToHtmlOptions DEFAULT = new DocumentToHtmlOptions("", false, StyleMap.EMPTY, false); - - private final String idPrefix; - private final boolean preserveEmptyParagraphs; - private final StyleMap styleMap; - private final boolean disableDefaultStyleMap; - - public DocumentToHtmlOptions(String idPrefix, boolean preserveEmptyParagraphs, StyleMap styleMap, boolean disableDefaultStyleMap) { - this.idPrefix = idPrefix; - this.preserveEmptyParagraphs = preserveEmptyParagraphs; - this.styleMap = styleMap; - this.disableDefaultStyleMap = disableDefaultStyleMap; - } - - public DocumentToHtmlOptions idPrefix(String prefix) { - return new DocumentToHtmlOptions(prefix, preserveEmptyParagraphs, styleMap, disableDefaultStyleMap); - } - - public DocumentToHtmlOptions preserveEmptyParagraphs() { - return new DocumentToHtmlOptions(idPrefix, true, styleMap, disableDefaultStyleMap); - } - - public DocumentToHtmlOptions addStyleMap(String styleMap) { - return addStyleMap(StyleMapParser.parse(styleMap)); - } - - public DocumentToHtmlOptions addStyleMap(StyleMap styleMap) { - return new DocumentToHtmlOptions(idPrefix, preserveEmptyParagraphs, this.styleMap.update(styleMap), disableDefaultStyleMap); - } - - public DocumentToHtmlOptions disableDefaultStyleMap() { - return new DocumentToHtmlOptions(idPrefix, preserveEmptyParagraphs, styleMap, true); - } - - public String idPrefix() { - return idPrefix; - } - - public boolean shouldPreserveEmptyParagraphs() { - return preserveEmptyParagraphs; - } - - public StyleMap styleMap() { - return disableDefaultStyleMap ? styleMap : DefaultStyles.DEFAULT_STYLE_MAP.update(styleMap); - } -} diff --git a/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java b/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java new file mode 100644 index 0000000..deab0d3 --- /dev/null +++ b/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java @@ -0,0 +1,249 @@ +package org.zwobble.mammoth.internal.conversion; + +import org.zwobble.mammoth.internal.documents.*; +import org.zwobble.mammoth.internal.html.Html; +import org.zwobble.mammoth.internal.html.HtmlNode; +import org.zwobble.mammoth.internal.results.InternalResult; +import org.zwobble.mammoth.internal.styles.HtmlPath; +import org.zwobble.mammoth.internal.styles.StyleMap; +import org.zwobble.mammoth.internal.util.Lists; + +import java.io.IOException; +import java.util.*; + +import static org.zwobble.mammoth.internal.util.Lists.*; +import static org.zwobble.mammoth.internal.util.Maps.map; +import static org.zwobble.mammoth.internal.util.Streams.toByteArray; + +public class DocumentToHtml { + public static InternalResult> convertToHtml(Document document, DocumentToHtmlOptions options) { + DocumentToHtml documentConverter = new DocumentToHtml(options); + return new InternalResult<>( + documentConverter.convertToHtml(document), + documentConverter.warnings); + } + + private static List findNotes(Document document, Iterable noteReferences) { + return eagerMap( + noteReferences, + // TODO: handle missing notes + reference -> document.getNotes().findNote(reference.getNoteType(), reference.getNoteId()).get()); + } + + public static InternalResult> convertToHtml(DocumentElement element, DocumentToHtmlOptions options) { + DocumentToHtml documentConverter = new DocumentToHtml(options); + return new InternalResult<>( + documentConverter.convertToHtml(element), + documentConverter.warnings); + } + + private final String idPrefix; + private final boolean preserveEmptyParagraphs; + private final StyleMap styleMap; + private final List noteReferences = new ArrayList<>(); + private final Set warnings = new HashSet<>(); + + private DocumentToHtml(DocumentToHtmlOptions options) { + this.idPrefix = options.idPrefix(); + this.preserveEmptyParagraphs = options.shouldPreserveEmptyParagraphs(); + this.styleMap = options.styleMap(); + } + + private List convertToHtml(Document document) { + List mainBody = convertChildrenToHtml(document); + // TODO: can you have note references inside a note? + List notes = findNotes(document, noteReferences); + if (notes.isEmpty()) { + return mainBody; + } else { + HtmlNode noteNode = Html.element("ol", + eagerMap(notes, this::convertToHtml)); + + return eagerConcat(mainBody, list(noteNode)); + } + } + + private HtmlNode convertToHtml(Note note) { + String id = generateNoteHtmlId(note.getNoteType(), note.getId()); + String referenceId = generateNoteRefHtmlId(note.getNoteType(), note.getId()); + List noteBody = convertToHtml(note.getBody()); + // TODO: we probably want this to collapse more eagerly than other collapsible elements + // -- for instance, any paragraph will probably do, regardless of attributes. (Possible other elements will do too.) + HtmlNode backLink = Html.collapsibleElement("p", list( + Html.text(" "), + Html.element("a", map("href", "#" + referenceId), list(Html.text("↑"))))); + return Html.element("li", map("id", id), eagerConcat(noteBody, list(backLink))); + } + + private List convertToHtml(List elements) { + return eagerFlatMap( + elements, + this::convertToHtml); + } + + private List convertChildrenToHtml(HasChildren element) { + return convertToHtml(element.getChildren()); + } + + private List convertToHtml(DocumentElement element) { + return element.accept(new DocumentElementVisitor>() { + @Override + public List visit(Paragraph paragraph) { + List content = convertChildrenToHtml(paragraph); + List children = preserveEmptyParagraphs ? cons(Html.FORCE_WRITE, content) : content; + HtmlPath mapping = styleMap.getParagraphHtmlPath(paragraph) + .orElseGet(() -> { + if (paragraph.getStyle().isPresent()) { + warnings.add("Unrecognised paragraph style: " + paragraph.getStyle().get().describe()); + } + return HtmlPath.element("p"); + }); + return mapping.wrap(children); + } + + @Override + public List visit(Run run) { + List nodes = convertChildrenToHtml(run); + if (run.isStrikethrough()) { + nodes = styleMap.getStrikethrough().orElse(HtmlPath.collapsibleElement("s")).wrap(nodes); + } + if (run.isUnderline()) { + nodes = styleMap.getUnderline().orElse(HtmlPath.EMPTY).wrap(nodes); + } + if (run.getVerticalAlignment() == VerticalAlignment.SUBSCRIPT) { + nodes = list(Html.collapsibleElement("sub", nodes)); + } + if (run.getVerticalAlignment() == VerticalAlignment.SUPERSCRIPT) { + nodes = list(Html.collapsibleElement("sup", nodes)); + } + if (run.isItalic()) { + nodes = styleMap.getItalic().orElse(HtmlPath.collapsibleElement("em")).wrap(nodes); + } + if (run.isBold()) { + nodes = styleMap.getBold().orElse(HtmlPath.collapsibleElement("strong")).wrap(nodes); + } + HtmlPath mapping = styleMap.getRunHtmlPath(run) + .orElseGet(() -> { + if (run.getStyle().isPresent()) { + warnings.add("Unrecognised run style: " + run.getStyle().get().describe()); + } + return HtmlPath.EMPTY; + }); + return mapping.wrap(nodes); + } + + @Override + public List visit(Text text) { + if (text.getValue().isEmpty()) { + return list(); + } else { + return list(Html.text(text.getValue())); + } + } + + @Override + public List visit(Tab tab) { + return list(Html.text("\t")); + } + + @Override + public List visit(LineBreak lineBreak) { + return list(Html.selfClosingElement("br")); + } + + @Override + public List visit(Table table) { + return list(Html.element("table", convertChildrenToHtml(table))); + } + + @Override + public List visit(TableRow tableRow) { + return list(Html.element("tr", convertChildrenToHtml(tableRow))); + } + + @Override + public List visit(TableCell tableCell) { + return list(Html.element("td", + Lists.cons(Html.FORCE_WRITE, convertChildrenToHtml(tableCell)))); + } + + @Override + public List visit(Hyperlink hyperlink) { + Map attributes = map("href", generateHref(hyperlink)); + return list(Html.collapsibleElement("a", attributes, convertChildrenToHtml(hyperlink))); + } + + private String generateHref(Hyperlink hyperlink) { + if (hyperlink.getHref().isPresent()) { + return hyperlink.getHref().get(); + } else if (hyperlink.getAnchor().isPresent()) { + return "#" + generateId(hyperlink.getAnchor().get()); + } else { + return ""; + } + } + + @Override + public List visit(Bookmark bookmark) { + return list(Html.element("a", map("id", generateId(bookmark.getName())), list(Html.FORCE_WRITE))); + } + + @Override + public List visit(NoteReference noteReference) { + noteReferences.add(noteReference); + String noteAnchor = generateNoteHtmlId(noteReference.getNoteType(), noteReference.getNoteId()); + String noteReferenceAnchor = generateNoteRefHtmlId(noteReference.getNoteType(), noteReference.getNoteId()); + return list(Html.element("sup", list( + Html.element("a", map("href", "#" + noteAnchor, "id", noteReferenceAnchor), list( + Html.text("[" + noteReferences.size() + "]")))))); + } + + @Override + public List visit(Image image) { + // TODO: custom image handlers + // TODO: handle empty content type + return image.getContentType() + .map(contentType -> { + try { + Map attributes = new HashMap<>(); + + String base64 = Base64.getEncoder().encodeToString(toByteArray(image.open())); + String src = "data:" + contentType + ";base64," + base64; + attributes.put("src", src); + + image.getAltText().ifPresent(altText -> attributes.put("alt", altText)); + + return list(Html.selfClosingElement("img", attributes)); + } catch (IOException exception) { + warnings.add(exception.getMessage()); + return Lists.list(); + } + }) + .orElse(list()); + } + }); + } + + private String generateNoteHtmlId(NoteType noteType, String noteId) { + return generateId(noteTypeToIdFragment(noteType) + "-" + noteId); + } + + private String generateNoteRefHtmlId(NoteType noteType, String noteId) { + return generateId(noteTypeToIdFragment(noteType) + "-ref-" + noteId); + } + + private String noteTypeToIdFragment(NoteType noteType) { + switch (noteType) { + case FOOTNOTE: + return "footnote"; + case ENDNOTE: + return "endnote"; + default: + throw new UnsupportedOperationException(); + } + } + + private String generateId(String bookmarkName) { + return idPrefix + bookmarkName; + } +} diff --git a/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtmlOptions.java b/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtmlOptions.java new file mode 100644 index 0000000..2de828d --- /dev/null +++ b/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtmlOptions.java @@ -0,0 +1,53 @@ +package org.zwobble.mammoth.internal.conversion; + +import org.zwobble.mammoth.internal.styles.DefaultStyles; +import org.zwobble.mammoth.internal.styles.StyleMap; +import org.zwobble.mammoth.internal.styles.parsing.StyleMapParser; + +public class DocumentToHtmlOptions { + public static final DocumentToHtmlOptions DEFAULT = new DocumentToHtmlOptions("", false, StyleMap.EMPTY, false); + + private final String idPrefix; + private final boolean preserveEmptyParagraphs; + private final StyleMap styleMap; + private final boolean disableDefaultStyleMap; + + public DocumentToHtmlOptions(String idPrefix, boolean preserveEmptyParagraphs, StyleMap styleMap, boolean disableDefaultStyleMap) { + this.idPrefix = idPrefix; + this.preserveEmptyParagraphs = preserveEmptyParagraphs; + this.styleMap = styleMap; + this.disableDefaultStyleMap = disableDefaultStyleMap; + } + + public DocumentToHtmlOptions idPrefix(String prefix) { + return new DocumentToHtmlOptions(prefix, preserveEmptyParagraphs, styleMap, disableDefaultStyleMap); + } + + public DocumentToHtmlOptions preserveEmptyParagraphs() { + return new DocumentToHtmlOptions(idPrefix, true, styleMap, disableDefaultStyleMap); + } + + public DocumentToHtmlOptions addStyleMap(String styleMap) { + return addStyleMap(StyleMapParser.parse(styleMap)); + } + + public DocumentToHtmlOptions addStyleMap(StyleMap styleMap) { + return new DocumentToHtmlOptions(idPrefix, preserveEmptyParagraphs, this.styleMap.update(styleMap), disableDefaultStyleMap); + } + + public DocumentToHtmlOptions disableDefaultStyleMap() { + return new DocumentToHtmlOptions(idPrefix, preserveEmptyParagraphs, styleMap, true); + } + + public String idPrefix() { + return idPrefix; + } + + public boolean shouldPreserveEmptyParagraphs() { + return preserveEmptyParagraphs; + } + + public StyleMap styleMap() { + return disableDefaultStyleMap ? styleMap : DefaultStyles.DEFAULT_STYLE_MAP.update(styleMap); + } +} -- cgit v1.2.3