diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/TODO | 3 | ||||
| -rw-r--r-- | src/builder.h | 83 |
2 files changed, 86 insertions, 0 deletions
@@ -6,3 +6,6 @@ that the memory is held by an auto_ptr until the constructor is done; otherwise, an exception raised from within the constructor will not call the destructor to free the memory. + +- Using mmap for the binary reader; or determine if the performance is + even worth the maintenance headaches of that code altogether. diff --git a/src/builder.h b/src/builder.h new file mode 100644 index 00000000..00adf483 --- /dev/null +++ b/src/builder.h @@ -0,0 +1,83 @@ +#ifndef _BUILDER_H +#define _BUILDER_H + +#include "xml.h" + +namespace ledger { + +/** + * @class builder_t + * + * @brief Represents an interface for building a data hierarchy. + * + * This interface is much like .NET's XmlWriter facility. It + * abstracts the kind of hierarchy we're building, instead focusing + * only on the relationships. + */ +class builder_t +{ + virtual void pushAttribute(const string& name, + const string& value) = 0; + virtual void pushAttribute(const nameid_t name_id, + const string& value) = 0; + + virtual void beginNode(const string& name) = 0; + virtual void beginNode(const nameid_t name_id) = 0; + + virtual void appendText(const string& text) = 0; + + virtual node_t * endNode(const optional<string>& name = + optional<string>()) = 0; + virtual node_t * endNode(const nameid_t name_id) = 0; +}; + +/** + * @class xml_builder_t + * + * @brief Build a generic node_t hierarchy. + * + * This builder can be used to parse ordinary XML into a document + * object structure which can then be traversed in memory. + */ +class xml_builder_t : public builder_t +{ +}; + +/** + * @class journal_builder_t + * + * @brief This custom builder creates an XML-mirrored Ledger journal. + * + * Rather than simply creating a node_t hierarchy, as xml_builder_t + * does, this code creates the associated journal elements referred to + * by those nodes, and then refers to those elements via minimalist + * "shadow nodes". + * + * Thus, after building a <transaction> element, the element itself + * will have no children, but instead will point to a transaction_t + * object. If later an XPath expression desires to traverse the + * <transaction> element, all of the appropriate child nodes will be + * constructed on the fly, as if they'd been created in the first + * place by a regular xml_builder_t. + */ +class journal_builder_t : public xml_builder_t +{ +}; + +/** + * @class xml_writer_t + * + * @brief Create textual XML on the given output stream. + * + * This builder, rather than manipulating data structures in memory, + * simply streams its contents on the fly to the given output stream. + * It uses only enough memory to remember the currently push + * attributes and text. + */ +class xml_writer_t : public builder_t +{ +}; + +} // namespace ledger + +#endif // _BUILDER_H |