source: docs/Working/icXML/arch-overview.tex @ 2866

Last change on this file since 2866 was 2866, checked in by nmedfort, 7 years ago


File size: 7.7 KB
3\def \CSG{Stream Generator}
5\icXML{} is more than an optimized version of Xerces. Many components were grouped, restructured and
6rearchitected with pipeline parallelism in mind.
7In this section, we highlight the core differences between the two systems.
8As shown in Figure \ref{fig:xerces-arch}, Xerces
9is comprised of five main modules: the transcoder, reader, scanner, namespace binder, and validator.
10The {\it Transcoder} converts source data into UTF-16 before Xerces parses it as XML;
11the majority of the character set encoding validation is performed as a byproduct of this process.
12The {\it Reader} is responsible for the streaming and buffering of all raw and transcoded (UTF-16) text.
13It tracks the current line/column position,
14%(which is reported in the unlikely event that the input contains an error),
15performs line-break normalization and validates context-specific character set issues,
16such as tokenization of qualified-names.
17The {\it Scanner} pulls data through the reader and constructs the intermediate representation (IR)
18of the document; it deals with all issues related to entity expansion, validates
19the XML well-formedness constraints and any character set encoding issues that cannot
20be completely handled by the reader or transcoder (e.g., surrogate characters, validation
21and normalization of character references, etc.)
22The {\it Namespace Binder} is a core piece of the element stack.
23It handles namespace scoping issues between different XML vocabularies.
24This allows the scanner to properly select the correct schema grammar structures.
25The {\it Validator} takes the IR produced by the Scanner (and
26potentially annotated by the Namespace Binder) and assesses whether the final output matches
27the user-defined DTD and schema grammar(s) before passing it to the end-user.
32\caption{Xerces Architecture} 
37In \icXML{} functions are grouped into logical components.
38As shown in Figure \ref{fig:icxml-arch}, two major categories exist: (1) the \PS{} and (2) the \MP{}.
39All tasks in (1) use the Parabix Framework \cite{HPCA2012}, which represents data as a set of parallel bit streams.
40The {\it Character Set Adapter}, discussed in Section \ref{arch:character-set-adapter},
41mirrors Xerces's Transcoder duties; however instead of producing UTF-16 it produces a
42set of lexical bit streams, similar to those shown in Figure \ref{fig:parabix1}.
43These lexical bit streams are later transformed into UTF-16 in the \CSG{},
44after additional processing is performed.
45The first precursor to producing UTF-16 is the {\it Parallel Markup Parser} phase.
46It takes the lexical streams and produces a set of marker bit streams in which a 1-bit identifies
47significant positions within the input data. One bit stream for each of the critical piece of information is created, such as
48the beginning and ending of start tags, end tags, element names, attribute names, attribute values and content.
49Intra-element well-formedness validation is performed as an artifact of this process.
50Like Xerces, \icXML{} must provide the Line and Column position of each error.
51The {\it Line-Column Tracker} uses the lexical information to keep track of the document position(s) through the use of an
52optimized population count algorithm, described in Section \ref{section:arch:errorhandling}.
53From here, two data-independent branches exist: the Symbol Pesolver and Content Preperation Unit.
55A typical XML file contains few unique element and attribute names---but each of them will occur frequently.
56\icXML{} stores these as distinct data structures, called symbols, each with their own global identifier (GID).
57Using the symbol marker streams produced by the Parallel Markup Parser, the {\it Symbol Resolver} scans through
58the raw data to produce a sequence of GIDs, called the {\it symbol stream}.
60The final components of the \PS{} are the {\it Content Preperation Unit} and {\it \CSG{}}.
61The former takes the (transposed) basis bit streams and selectively filters them, according to the
62information provided by the Parallel Markup Parser, and the latter transforms the
63filtered streams into the tagged UTF-16 {\it content stream}, discussed in Section \ref{section:arch:contentstream}.
65Combined, the symbol and content stream form \icXML{}'s compressed IR of the XML document.
66The {\it \MP{}}~parses the IR to validate and produce the sequential output for the end user.
67The {\it Final WF checker} performs inter-element wellformedness validation that would be too costly
68to perform in bitspace, such as ensuring every start tag has a matching end tag.
69Xerces's namespace binding functionality is replaced by the {\it Namespace Processor}. Unlike Xerces,
70it is a discrete phase that produces a series of URI identifiers (URI IDs), the {\it URI stream}, which are
71associated with each symbol occurrence.
72This is discussed in Section \ref{section:arch:namespacehandling}.
73Finally, the {\it Validation} layer implements the Xerces's validator.
74However, preprocessing associated with each symbol greatly reduces the work of this stage.
80\caption{\icXML{} Architecture}
85% Probably not the right area but should we discuss issues with Xerces design that we tried to correct?
86% - over-reliance on hash tables when domain knowledge dictated none would be needed
87% - constant buffering of text to ensure that every QName/NCName and content was contained within a single string
88% - abundant use of heap allocated memory
89% - text conversions done in multiple areas
90% - poor cache utilization; attempted to improve by using smaller layers of tasks in bulk
92% As the previous section aluded, the greatest difference between sequential parsing methods
93% and the Parabix parsing model is how data is processed.
94% Consider Figure \ref{fig:parabix1} again. In it, the start tags are located independent of the end
95% tags. In order to produce Xerces-equivalent output, icXML must emit the start and end tag
96% events in sequential order, with all attribute data associated with the correct tag.
100% The Parabix framework, however, does not allow for this (and would be hindered performance wise if
101% forced to.)
102% Thus our first question was, ``How can we how can we take full advantage
103% of Parabix whilst producing Xerces-equivalent output?'' Our answer came by analyzing what Xerces produced
104% when given an input text.
106% By analyzing Xerces internal data structures and its produced output, two major observations were obvious:
107% (1) input data is transcoded into UTF-16 to ensure that there is a single standard character type, both
108% internally (within the grammar structures and hash tables) and externally (for the end user).
109% (2) all elements and attributes (both qualified and unqualified) are associated with a unique element
110% declaration or attribute definition within a specific grammar structure. Xerces emits the appropriate
111% grammar reference in place of the element or attribute string.
117%   From Xerces to icXML
119%   - Philosophy:  Maximizing Bit Stream Processing
121%   - Character Set Adapters vs. Transcoding
122%   - Bitstreams 1: Charset Validation and Transcoding equations
123%   - Bitstreams 2: Parabix style parsing and validation
125%   - Bitstreams 3: Parallel filtering and normalization
126%           - LB normalization
127%           - reference compression -> single code unit speculation
128%           - parallel string termination
130%   - Bitstreams 4: Symbol processing
132%   - From bit streams to doublebyte streams: the content buffer
134%   - Namespace Processing: A Bitset approach.
Note: See TracBrowser for help on using the repository browser.