X-Git-Url: https://dreyeck.freedombox.rocks/gitweb/topic-maps.git/blobdiff_plain/5e716c078b885e97ebe858230ce7d8578e65ae4b..refs/heads/codex/xtm-slice-22r:/src/ctm/reader.lisp?ds=sidebyside diff --git a/src/ctm/reader.lisp b/src/ctm/reader.lisp index c510c63..f8741a7 100644 --- a/src/ctm/reader.lisp +++ b/src/ctm/reader.lisp @@ -2,7 +2,7 @@ (defvar *active-resource-iris* nil) -(defstruct template-definition name parameters body position) +(defstruct template-definition name parameters body position source-name source-text document-iri) ;; Validated concrete syntax, never TMDM information items or Lisp forms. (defstruct ctm-statement kind tokens position) (defstruct template-argument kind token topic value datatype plain-string-p) @@ -180,9 +180,10 @@ The owning information item is not constructed until this function finishes." :datetime-bce-timezone-rollover :datetime-leap-second-context) (token-position token) "DateTime normalization is outside the supported BCE/leap-second calendar context.")) (values value (topic-maps.tmdm:datetime-datatype)))) - (:excluded-literal - (problem 'ctm-unsupported (if (eq context :variant) :variant-literal-datatype :occurrence-literal-datatype) (token-position token) - "The separate CTM '*' literal is outside this slice.")) + (:ctm-integer + (consume parser) + ;; CTM 3.4.1 [33]: opaque lexical value, unrelated to Wildcards/counters. + (values "*" "http://psi.topicmaps.org/iso13250/ctm-integer")) (:string (consume parser) (if (punctuation-p parser #\^) @@ -415,10 +416,38 @@ Use the shared Topic reference resolver." (attach-reifier role reifier marker)))) (attach-reifier (topic-maps.tmdm:canonical-association association) r position)))))) -(defun deserialize-string (text &key document-iri source-name (wildcard-iri document-iri) (wildcard-counter 0) +(defun read-prolog (parser) + "CTM 3.5: one initial encoding header, then one optional version 1.0." + (when (and (eq :directive (token-kind (look parser))) + (string= "encoding" (token-value (look parser)))) + (let ((token (consume parser))) + (unless (zerop (token-position token)) + (problem 'ctm-rejection :misplaced-encoding (token-position token) + "Encoding declaration must be the first characters of the CTM stream.")) + (multiple-value-bind (name end position) + (ctm-encoding-name *source-text* (token-position token) :character) + (unless (ctm-encoding-kind name) + (problem 'ctm-unsupported :unsupported-encoding position "Only UTF-8 and ISO-8859-5 CTM input are implemented.")) + (setf (lexer-cursor (parser-lexer parser)) end)))) + (when (and (eq :directive (token-kind (look parser))) + (string= "version" (token-value (look parser)))) + (consume parser) + (let ((lexer (parser-lexer parser))) + (skip-layout lexer) + (let ((start (lexer-cursor lexer))) + ;; Match the fixed terminal before the numeric-literal lexer. A Topic + ;; can follow immediately; extra numeric spelling is not version 1.0. + (unless (and (ctm-ascii-prefix-p *source-text* start "1.0") + (let ((end (+ start 3))) + (not (and (< end (length *source-text*)) + (find (char *source-text* end) "0123456789.+-"))))) + (problem 'ctm-rejection :invalid-version start "The only CTM version is the exact spelling 1.0.")) + (setf (lexer-cursor lexer) (+ start 3)))))) + +(defun deserialize-context (text &key document-iri source-name (wildcard-iri document-iri) (wildcard-counter 0) resource-resolver (max-resource-depth 32)) "Read the documented CTM subset, with per-document prefixes and qualified strings. -Return the Map and final Wildcard Counter as two values. Wildcard IRI defaults +Internal result: Map, final Wildcard Counter and parser environment. Wildcard IRI defaults to Document IRI; local identifiers and relative IRIs still use Document IRI. Source positions are zero-based character offsets; lines/columns are one-based. RESOURCE-RESOLVER, if supplied, is called with an absolute IRI and :PURPOSE, @@ -446,6 +475,7 @@ There is no implicit retrieval. MAX-RESOURCE-DEPTH bounds nested imports (defaul (topic-maps.tmdm:core-subject-error (lambda (c) (problem 'ctm-rejection (topic-maps.tmdm:core-subject-error-code c) (lexer-cursor (parser-lexer parser)) (princ-to-string c))))) + (read-prolog parser) (loop while (eq :directive (token-kind (look parser))) do (read-directive parser map)) (multiple-value-bind (r position) (read-reifier parser map) (attach-reifier map r position)) (loop until (eq :eof (token-kind (look parser))) do @@ -455,12 +485,26 @@ There is no implicit retrieval. MAX-RESOURCE-DEPTH bounds nested imports (defaul (cond ((eq :directive (token-kind (look parser))) (read-directive parser map)) ((identifier-p (look parser) "def") (read-template-definition parser)) (t (read-statement parser map))))) - (values map (parser-wildcard-counter parser))))) + (values map (parser-wildcard-counter parser) parser)))) + +(defun deserialize-string (text &key document-iri source-name (wildcard-iri document-iri) (wildcard-counter 0) + resource-resolver (max-resource-depth 32)) + "Deserialize CTM using explicit Document IRI and optional controlled resources. +Return the Topic Map and final Wildcard Counter. Includes import Templates and +Topic aliases; Mergemap imports Map content only. No implicit I/O is performed. +The resolver receives :PURPOSE (:INCLUDE or :MERGEMAP), :FROM-DOCUMENT-IRI and +:NOTATION (NIL for Include). Source keys are diagnostic, never identity bases." + (multiple-value-bind (map counter) + (deserialize-context text :document-iri document-iri :source-name source-name + :wildcard-iri wildcard-iri :wildcard-counter wildcard-counter + :resource-resolver resource-resolver :max-resource-depth max-resource-depth) + (values map counter))) -(defun deserialize-file (pathname &key document-iri source-name (wildcard-iri document-iri) (wildcard-counter 0) +(defun deserialize-file (pathname &key document-iri source-name encoding (wildcard-iri document-iri) (wildcard-counter 0) resource-resolver (max-resource-depth 32)) - "Read a UTF-8 CTM file, with optional UTF-8 BOM, using an explicit Document IRI. -Other encoding directives are UNSUPPORTED. File I/O errors propagate unchanged." + "Read CTM octets using an explicit Document IRI and optional supplied ENCODING. +Check supplied encoding/BOM/declaration conflicts before UTF-8 or ISO-8859-5 decoding. +Other charsets remain UNSUPPORTED. File I/O errors propagate unchanged." (let ((*source-name* source-name) (*source-text* "")) (require-document-iri document-iri) (require-wildcard-environment wildcard-iri wildcard-counter) @@ -468,7 +512,7 @@ Other encoding directives are UNSUPPORTED. File I/O errors propagate unchanged." (let ((bytes (make-array (file-length input) :element-type '(unsigned-byte 8)))) (unless (= (length bytes) (read-sequence bytes input)) (error "Input file changed or was truncated during reading.")) - (deserialize-string (decode-utf8 bytes :strip-bom t) + (deserialize-string (decode-ctm-octets bytes :encoding encoding) :document-iri document-iri :source-name source-name :wildcard-iri wildcard-iri :wildcard-counter wildcard-counter :resource-resolver resource-resolver :max-resource-depth max-resource-depth)))))