RASMUS User`s Manual
Contents
1. Exp not Exp Exp and Exp Exp or Exp Exp Exp Exp Exp Exp 10 AtomConst BoolConst IntConst Digit Text Const Exp Exp Exp Exp Exp mod Exp Exp Exp Exp Exp Exp Exp Exp Exp Exp lt lt Exp Exp Name Exp Name has Exp Name Exp ProjSym Name A Exp Exp Exp RenamePair A 7 Exp Restrict Exp I lt Exp Restrict Exp I gt Exp Restrict Exp max Exp Name min Exp Name count Exp Name add Exp Name mult Exp Name Exp Exp Exp lt gt Exp Exp Exp Exp gt Exp Exp lt Exp Exp gt Exp Exp Exp if Guards fi NameValue in Exp Exp Exp Exp BoolConst IntConst TextConst true false Digit 0 1 2 3 4 5 6 7 8 9 Ascii 11 Ascii Letter SpecialChar NameType Name AlphaNum Type StandardConst RelConst NameExp ProjSym RenamePair Restrict Guards Choice NameValue IsType Letter Digit SpecialChar a b 1 1e s28 1 27l Sd sed Ae p pede pe ad Name Type Letter AlphaNum Letter Digit Bool Int Text Tup Rel Func Any Bool Int Text zero one Name Exp I l Name lt Name Name A Exp gt Exp Choice amp Guards val Name Exp is Bool Exp is Int Exp is Text Exp is Tup2. Delete relation X This will irretrievably delete the relation YoungPeople Cancel OK Figure 6 Deleting a relation number of attribute names type name type name field field Figure 7 External relation format 2 T Name I Age Bruce Jones 5 Mary Ross 12 Ann Bird 12 Kenneth Lewis 17 Figure 8 External relation format of the children relation ne E ae Figure 9 External CSV format Name Age Bruce Jones 5 Mary Rose 12 Ann Bird 12 Kenneth Lewis 17 Figure 10 The children relation as a CSV file 4 Expressions In this chapter we define the set of RASMUS expressions We do this in several steps First we give a grammar from which expressions must be generated Af terwards we give an informal semantics of RASMUS expressions In connection with this semantics we restrict the set of expressions further by adding type constraints 4 1 Grammar In the following we define some notation e Category is a sequence of Category Category is a nonempty sequence of Category A Category is a comma separated sequence of Category Category is a nonempty comma separated sequence of Category Category is either 0 or 1 of Category In the following a grammar for RASMUS expressions is presented Exp AtomConst RelConst StandardConst tup NameExp rel Exp func NameType gt Type Exp end Name Name Exp
3. ri A1 An denotes the projection of r onto the attributes in the schema of r except the attributes A1 An r b where b is a boolean expression contains the tuples t from r which make b true when the special symbol is replaced with t r A1 B1 An Bn denotes a renaming of the attributes in r The attribute names A1 An are changed to B1 Bn respectively An error occurs if the attributes A1 An do not belong to the schema of r The Ai s must be pairwise different Also the Bi s must be pairwise different and no Bi is allowed to belong the schema of r minus A1 An r1 rn X exp denotes a factor expression The result of a factor expression is the union the evaluation of a family of expressions to be described in the following These must all evaluate to relations with the same schema Otherwise an error occurs The family of expressions is constructed by taking exp and substituting 4 0 1 0 n with different tuple and relation values The values for 0 1 6 n are determined as follows X must be a comma separated list of the attributes in the intersection of the schemas of the relations ri through rn The result of evaluating ri X rn X is called the base relation The symbol is instantiated with the tuples in the base relation one at a time Now assume that has been given a value then 7 is the relation rel G ri X If X is not specified then it is assumed to b
4. is equal to t 0 tI Standard values The atomic types have standard values These are written Bool Int and Text These values are outside the ordering This means for example that if 14 i is not the standard value Int then the expressions i lt Int i Int and i gt Int will all evaluate to Bool Similarly if we do arithmetic with Int the result will also be a Int Tuples A tuple is a set of pairs where each pair consists of an attribute name and an atomic value If A1 An are attribute names and v1 vn are atomic values then a tuple can be specified as tup A1 v1 An vn We have the following operations on tuples e t1 lt lt t2 denotes the tuple t1 updated with the tuple t2 The expression t1 t2 basically evaluates to the union of t1 and t2 except that whenever an attribute name appears in both t1 and t2 only the attribute name and corresponding value from t2 is used If the same attribute name appears in both arguments but the values are of different types then an error occurs e t A denotes the value associated with the attribute name A in t If A does not appear in t then an error will occur e t A denotes the tuple t except that the attribute name A and its associated value is left out If A does not appear in t then an error will occur e has t A denotes the boolean value true if the attribute name A appears in t If not then the value false is returned Relations A rel
5. Aarhus Universitet Institut for Datalogi dPersp14 RASMUS User s Manual Contents 1 Introduction 1 2 The Interpreter 1 2 1 Getting started 2 eee 1 2 2 Getting along uu io Pe ON Ed ed CRUS ie RU 2 2 9 The Editor 22 tote titel Qs Bde BP ES Ae wd 3 2 4 Preferences and Help o o 4 3 Persistence 4 3 1 Starting and Ending a Session o 4 3 2 External Relation Format 24 4 5 3 3 Working with CSV Files o o 7 4 Expressions 10 TT Grammars si e ea e aa qd X AY ue 10 4 2 JIKeyWOEPGS y Sus metals m utin A Poss My GU E I ue qu s 13 4 3 InformalSemantics 22e 13 File Edit Help Name Value Kampe Relation Rundel Relation Runde2 Relation Figure 1 The initial RASMUS window 1 Introduction This manual describes how to use the Rasmus system It also describes how to obtain persistence i e how to save results between sessions Finally it describes what Rasmus expressions look like and what they mean 2 The Interpreter The purpose of the Rasmus interpreter is to evaluate expressions typed by the user This chapter describes the RASMUS interpreter 2 1 Getting started To start RASMUS execute the rasmus file This opens a graphical user interface GUI window similar to the one shown in figure 1 In the following we describe how to evaluate RASMUS expressions in the inter Kampe Relation File He
6. Exp is Rel Exp is Func Exp is Any Exp is Bool Exp Name is Int Exp Name is Text Exp Name 12 4 2 Keywords A Name cannot be one of the keywords listed in the following add and any bool count end false fi func has if in int max min mod mult not one or rel text true tup unset val zero For these keywords the system does not distinguish between upper and lower case letters 4 3 Informal Semantics We divide this section up into several subsections depending on the type of the expressions being discussed Some operations involve more than one type but are only discussed once We try to place such operations under the main type involved First we describe the atomic values booleans integers and text Booleans The constants true and false along with the operations not and and or have the standard interpretations They require boolean arguments and they return a boolean value Values of the same type can be compared and the result of a comparison is a boolean Any two values can be compared using or lt gt Values of type Bool Int Text Tup and Rel can also be compared using lt gt lt and gt In following we list the results of comparing types using the test lt The remaining test have the obvious complementary interpretation Bool x lt y is true iff x is false and y is true 13 Int x lt y is true iff x is an integer less than y Text
7. al parameters The result of the function application is the value of the function body unless this value is not of the type specified in the function definition the result type e exp denotes whatever the expression exp denotes e is Bool exp denotes true if exp evaluates to a boolean Otherwise it denotes false The constructions is Int is Text is Atom is Tup is Rel is Func and is Any are defined similarly For the atomic types there is an additional construction If exp eval uates to a relation and A is an attribute name of that relation then is Bool exp A denotes true if A is of type Bool and false otherwise If either exp does not evaluate to a relation or exp evaluates to a rela tion and A does not belong to the schema of that relation then a runtime error occurs The expressions is Int exp A and is Text exp A have similar semantics e unset name unsets the contents of the variable name If name is a relation the corresponding relation file is permanently deleted 18 List of Figures 10 The initial RASMUS window Inspecting the relation Kampe The first RASMUS expression o e e SLIM Up 3 predea e a a be dudes Relation view of the example relation children Deleting a relation cata ia e d RI A AR EYE External relation format External relation format of the children relation External CSV format The children relation as CSV fil
8. ation Relation Relation 55 Figure 3 The first RASMUS expression Your RASMUS window will now look as in figure 3 Note that due to the last on the first line the interpreter was still waiting for input Had the been omitted the first line would have been evaluated before the second line could be typed in the interpreter Furthermore observe that a new name appeared in the environment list on the right This happened because a new variable x has been defined If you type an illegal expression the RASMUS interpreter will do its best to diagnose the error and provide a use the error What constitutes a legal e To close the RASMUS system go to File 2 3 The Editor and Quit ful error message such that you can correct xpression is further described in Chapter 4 Often we want to produce more complicated pieces of code rather than simple expressions To accomodate this the RASMUS system contains a development environment as well To open a new file in the editor go to File click New and then This opens an editor where RASMUS code can be written Note that there is simple syntax highlighting and intellisense i e a red lines appear under invalid expressions Once you have written your code you can send it to the interpreter by either going to File and click Run or by us
9. ation is a set of tuples such that every tuple contains the same set of attribute names and such that for any two tuples the values associated with identical attribute names are of the same type This set of attribute names with their associated types is called the schema of the relation If t is a tuple then rel t is a relation with one tuple t There are the following operations on relations e r denotes the length of the relation r i e the number of tuples in r 15 has r A denotes the boolean value true if the attribute name A appears in the schema of r If not then the value false is returned r1 r2 denotes the union of the tuples in r1 and r2 If r1 and r2 does not have the same schema then an error will occur ri r2 denotes the set difference of r1 and r2 ie ri r2 is the set of tuples from r1 which do not belong to r2 If r1 and r2 does not have the same schema then an error will occur ri r2 denotes the join of ri and r2 The attribute names appearing in both schemas are required to be of the same type Otherwise an error will occur The relation ri r2 consists of all the tuples t such that t restricted to the schema of r1 belongs to r1 and such that t restricted to the schema of r2 belongs to r2 ri A1 Andenotes the projection of r onto the attributes A1 An These attributes have to belong to the schema of r1 The relation consists of all the tuples from r1 restricted to the attributes A1 An
10. d when we ended the last session We discuss this further in section 3 1 Another difference is that a database should be able to deal with huge amounts of external data In particular a database language is required to include a specification of how external data can be read by the system This is the subject of section 3 2 and section 3 3 3 1 Starting and Ending a Session As described in the beginning RASMUS is started executing the rasmus file When starting a session RASMUS reads the Environment Path as set in Edit tory These relations are the files in that directory with the rdb file extension Preferences and establishes the relations represented by the direc As an example suppose we have started RASMUS with the Environment Path pointing to a directory where there is a relation called children After starting up the relation children is loaded and can be seen in the environment figure 4 Double clicking the children relation in the environment list brings up the relation view figure 5 Tf you want to export a relation from the RASMUS system go to File in the relation view and click Export CSV This brings up a dialog where you can decide where to save the file on your disk It is also possible to save copy relations under a different name by clicking the Save as global
11. e 19
12. e all the common attributes of the r s If a list of attributes is specified a restriction list then X is this list An error occurs if X is not contained in the intersection of the schemas of the n relational arguments 16 e The variants lt and gt have the same semantics except that the tu ples are processed in a non decreasing respectively non increasing order according to the attributes X e zero denotes the empty relation with the empty schema e one denotes the relation with the empty schema containing one tuple the empty tuple Aggregation The five operations max min count add and mult are very similar They are all used like max r A where r is a relation and A an attribute name They perform the action indicated by their name e g max r A returns the maximal value in the A column of the relation r An error occurs if the relation does not have an A column The standard values are always ignored If the A column contains nothing but standard values or if the relation is empty then max and min returns the standard value count and add returns 0 and mult returns 1 Miscellaneous e if bi gt exp1 amp amp bn gt expn fi is the conditional expression The bi s are boolean expression This conditional expression is evaluated as follows The boolean expressions are evaluated in order until one is found which gives true If none of the boolean expressions evaluate to true then the result is 0 If bi wa
13. ing the shortcut Ctrl R The code you type can be saved in RASMUS code files by using File Save As a side note RASMUS code files have the extension rm If you want to load previously written RASMUS code files you have to use the File menu in the main window followed by clicking Open and locating the file on your system 2 4 Preferences and Help It is possible to do some configuration of your RASMUS system such as chang ing font types and colors Going to Edit and Preferences opens a dia log where you can change these visual features An important setting is the Environment Path This is the directory where all relations are stored and also where they are automatically loaded from when RASMUS is started It is recommended that you use the Environment Path as workspace but you re member to export relations to other places as well Further information on how relations are stored and retrieved can be found in section 3 In the Help menu you can find a tutorial on RASMUS and the RASMUS manual this document 3 Persistence One of the primary differences between an ordinary programming language and a database language is that the latter works on persistent data i e we want to be able to start a session with the data we ha
14. into the system r RASMUS eo 88 File Edit Help Name Value children Relation Figure 4 Starting up File Edit Help Name Value Relation of 4 tuples mm children Relation t O X File Help Name Tex A Age Int Bruce Jones 5 Mary Rose 12 Ann Bird 12 Kenneth Lewis 17 Figure 5 Relation view of the example relation children A relation in external format looks as illustrated in figure 7 Here type can be either T I or B representing the three atomic types TEXT INTEGER and BOOLEAN The external format of the children relation would be as depicted in figure 8 Relations are stored in external format as ordinary text files The text file is named after the relation name by adding an extension of rdb So our example relation was stored in a file children rdb with the exact contents of figure 8 This file can be loaded into RASMUS by storing it in the Environment Path 3 3 Working with CSV Files Today much data is stored in CSV Comma Seperated Value files and RASMUS is capable of reading and writing relations from and to CSV files Here we describe how RASMUS interprets a CSV file when it is loaded A CSV files consists of zero or more lines Each line contains the same number of fields seperated commas A field may be enclosed by quotes lt content gt Let m be the number of fields and n be the number of rows then the content of the first field of the first
15. item in the File menu If you want to save the result of an earlier evaluated expression you must reevaluate it and assign it to a variable using the RASMUS assignment syntax in the interpreter Remember that the expression is listed somewhere in the interpreter s history and using the up and down arrow keys on your keyboard you go through the history of evaluated expressions Deleting a relation from the environment is achieved through the File menu in the main window and clicking Delete relation This brings up a dialog asking for the name of the relation to delete Suppose we had made a copy of the children relation by the name of youngpeople and we wanted to delete it Going through the two steps would bring us to the dialog seen in figure 6 Clicking Ok removes YoungPeople from the environment and it deletes the corresponding relation file in the Environment Path Note that a relation is irretrieveably lost when it is deleted 3 2 External Relation Format Sometimes data is produced by other programs or have been collected by people over time and the question naturally arises how can such data be read into a relation automatically For this purpose we describe the external relation format This is the format in which relations are kept in the directories So obviously if we can transform data to this format automatically then this data can be read
16. lp ld Int A HjHold Text UdeHold Text E 1 o AGF Brendby 2 Frem OB 3 AaB Lyngby 4 OB AGF 5 Brendby AaB 6 Lyngby Frem 7 AaB AGF 8 Frem Br ndby 9 Lyngby OB 10 AGF Frem Figure 2 Inspecting the relation Kampe Insepecting the relation Kampe Note that the relation cannot be modified here that is only possible through interaction with the RASMUS interpreter It is however possible to sort the tuples by clicking on the header Note that it is a stable sort so you can sort the tuples breaking ties properly preter Next we walk through the features that are available in the GUI 2 2 Getting along On the left side of the main window there is a Read Eval Print Loop REPL interpreter where RASMUS expressions can be typed and evaluated The left side will be refered to as the interpreter On the right side we see a list of names and values The names are the variables that currently exist in the global scope and the values are the corresponding values If a variable s value is Relation then it is possible to inspect that relation by double clicking the variable name This will pop up a new window similar to the one in figure 5 Typing in expressions in the interpreter and pressing return makes the inter preter evaluate the expression Suppose you have typed in the following expres sion X 1 2 3 4 5 6 7 8 9 10 LJ File Edit Help Name Kampe Rundel Runde2 Value Rel
17. row is f 1 and the content of the very last field is Fam When reading a CSV file RASMUS tries to guess the type of each attribute in the tuples of the relation This is done by guessing the type of each attribute in turn The type of the first attribute is inferred based on the content of the fields f21 f3 1 fn 1 In general the jth attribute s type is guessed based on the contents of f2 f3 j fn The type of an attribute is guessed as follows If all the fields contain an integer the type is INT If all the fields contain either true or false the type is BOOLEAN Otherwise the type is TEXT Note the first line of fields have been ignored This is because it might be the names of the attributes also known as the header of the column If all fields fi for 1 lt i n and 1 j m are textual fields then the first row is interpreted as values rather than as the names of the attributes In this case the names of the columns are column0 columni etc Otherwise the field fi for 1 lt j lt m is interpreted as the name of column j A relation in CSV format looks as illustrated in figure 9 The CSV file of the children relation would be as depicted in figure 10 Looking through figure 10 we see RASMUS can infer that the type of the Age column is INT since all values except the first value in that column are integers File Edit Help Name Value YoungPeople Relation of 4 tuples children Relation of 4 tuples
18. s the first expression evaluating to true then the result of the conditional expressions is the result of evaluating expi e val xi expi val xn expn in exp isa block The expressions exp1 through expn are evaluated in order and the results are named x1 through xn respectively Then exp is evaluated and returned as the result of the block Note that the values are named as soon as they are evaluated so the names x1 through x i 1 can be used in expi and all the names can be used in exp e expli exp2 is a sequence which evaluates first exp1 and then exp2 the result is that of exp2 e n exp is an assignment which evaluates exp and assigns the result to the identifier n the result is that of exp e func x1 T1 xn Tn T exp end is a function definition and denotes a function The names x1 through xn are the formal parameters and T T1 Tn are types T is called the result type 17 e f exp1 expn is a function application The function f must be defined in the environment The expressions exp1 through expn are eval uated in order and the values are bound to the formal parameters of the function definition An error occurs if the number of expressions does not equal the number of formal parameters An error also occurs if an expres sion evaluates to a value of type different from the one specified in the function definition The body of the function definition is now evaluated and it can depend on the form
19. x lt y is true iff x is a genuine prefix of y Tup x lt y is true iff the set of names in x is strictly contained in the set of names in y In addition the intersection of names in x and y should have the same associated values Rel x lt y is true iff the set of tuples in x is strictly contained in the set of tuples in y An error occurs if the schemas of x and y are different The comparison x y on texts holds true if x occurs as a subtext of y Integers The integer constants along with the operations and mod have the stan dard interpretations Only Int values in the range 2147483647 to 2147483647 are available A system error will occur if operations evaluate to values outside that range The operations listed above require integer arguments and they return an integer value We point out that the value of x y is the integer part of x divided by y and x mod y is the remainder of the same operation The expression i gives the same value as 0 1 Text A Text is a sequence of characters A constant text is written as a sequence of characters surrounded by double quotes i e like Rasmus e ti t2 denotes the concatenation of the texts t1 and t2 e t i j denotes the subsequence from t starting with the ith character and ending with the j 1 th character where i and j are integers A character sequence of length n is numbered from 0 to n 1 e t denotes the length of the text t For example any text t
Download Pdf Manuals
Related Search